Artificial intelligence vs. augmented intelligence Artificial intelligence vs. augmented intelligence

The AMA House of Delegates uses the term augmented intelligence (AI) as a conceptualization of artificial intelligence that focuses on AI’s assistive role, emphasizing that its design enhances human intelligence rather than replaces it.

AMA policy on AI development, deployment and use AMA policy on AI development, deployment and use

The AMA is committed to ensuring that AI can meet its full potential to advance clinical care and improve clinician well-being. As the number of AI-enabled health care tools continue to grow, it is critical they are designed, developed and deployed in a manner that is ethical, equitable and responsible. The use of AI in health care must be transparent to both physicians and patients.

In addition to medical devices, AI is increasingly used in health care administration or to reduce physician burden, and policy and guidance for both device and non-device use of health care AI is necessary. Recognizing this, the AMA has developed new policy (PDF) that addresses the development, deployment and use of health care AI, with particular emphasis on:

Health care AI oversight

When and what to disclose to advance AI transparency

Generative AI policies and governance

Physician liability for use of AI-enabled technologies

AI data privacy and cybersecurity

Payor use of AI and automated decision-making systems

Physician sentiments on AI Physician sentiments on AI

In 2023, the AMA conducted a comprehensive study of over 1,000 physicians’ sentiments towards the use of AI in health care including current use and future motivations for use, key concerns, areas of greatest opportunity and requirements for adoption.​ Given the rapidly evolving AI landscape across health care, the AMA repeated the study in late 2024 (PDF). The objectives of this research remain:

Capturing the sentiment among practicing physicians regarding the increased usage of AI in health care​

Evaluating AI use cases based on their familiarity, relevance, and usefulness​

Identifying key resources and areas of need for physicians to consider implementation of AI tools to their practice

Physicians largely remain enthusiastic about the potential of AI in health care, with 68% seeing at least some advantage to the use of AI in their practice, up from 65% in 2023. We also saw use of AI increase from 38% in 2023 to 66% of physicians reporting they use some type of AI tool in practice in 2024.

However, there are still key concerns as physicians continue to explore how these tools will impact their practices. Implementation guidance and research, including clinical evidence, remain critical to helping physicians adopt AI tools.

Physician sentiments study on AI AMA’s latest study on physician sentiments around the use of AI in heath care: motivations, opportunities, risks and use cases. Read Now (PDF)

AI in medical education AI in medical education

AI is playing an increasingly important role at all stages of the medical education continuum, both as a tool for educators and learners and as a subject of study in and of itself. AI has the potential to transform the educational experience as a part of precision education and transform patient care as a part of precision health. Learn more about how AI can impact medical education.

AI in health care CME AMA ChangeMedEd® Artificial Intelligence in Health Care Series helps learners understand the strengths and limitations of AI in health care and health systems. Start on AMA Ed Hub™

AMA partners with technology and health care leaders to bring physicians critical insights on AI’s potential applications and ensure that physicians have a voice in shaping AI’s role in medicine.

An AMA issue brief (PDF) provides a brief overview of recent state legislative activity and discusses three key AI policy areas for state legislative/regulatory activity: health plan use of AI, transparency and physician liability.

To develop actionable guidance for AI in health care, the AMA reviewed literature on the challenges health care AI poses and reflected on existing guidance. These findings are published in a new paper in Journal of Medical Systems: Trustworthy Augmented Intelligence in Health Care.

The AMA Intelligent Platform’s CPT® Developer Program allows developers to access the latest content and resources. Access the Developer Portal on the AMA Intelligent Platform.

Kimberly Lomis, MD, AMA vice president of undergraduate medical innovations, co-authors a discussion paper on Artificial Intelligence for Health Professions Educators in NAM Perspectives.

CPT® and AI CPT® and AI

The current CPT® code set drives communication across health care by enabling the seamless processing and advanced analytics for medical procedures and services.

AMA offers several resources to provide guidance on the updated CPT® code set for classifying various AI applications as well as advisory expertise through the Digital Medicine Payment Advisory Group (DMPAG). DMPAG identifies barriers to digital medicine adoption and proposes comprehensive solutions on coding, payment, coverage and more.

Stay up-to-date on the criteria for CPT® codes, access applications and read frequently asked questions.

Health care AI news Health care AI news

Learn how AI is being used in health care as the medical community’s understanding of the application grows. AMA articles focus on ways coding content advances to reflect the emergence of digital health and diagnostics, and how AI should be incorporated into physician training.

Stay up-to-date on information about health care AI, including the latest news, trends and AMA statements.

AI and practice management AI and practice management

The technological capacity exists for AI algorithms and tools to transform health care, but real challenges remain in ensuring that tools are developed, implemented and maintained responsibly in your practice. Learn more with the AMA.

Explore the current landscape of AI in medicine from terminology to current and future use cases to addressing risks. Download the report now.

AMA Board reports and policy AMA Board reports and policy

AMA Board of Trustees is responsible for implementing AMA policy. Given the number of stakeholders and policymakers involved in the evolution of AI in health care, it is important that AMA not only adopt a base level of policy to guide engagement but equally continue to refine policy as this technology develops.

AMA Board reports included here summarize the need for additional AMA policy on AI. Through the AMA PolicyFinder, users can search for current AI policy initiatives.

AI learning on Ed Hub and JAMA Network™ AI learning on Ed Hub and JAMA Network™

Explore the components of AI in health care and delve into the potential challenges and opportunities for physicians. An AMA Ed Hub™ article explores how AI, used ethically, has the power to serve as a transformative and powerful tool for physicians.

Healthcare is a data-rich industry ripe for artificial intelligence deployment. The impact of AI in healthcare has been positive.

Hospitals, health systems and other provider-based organizations have integrated AI into their daily workflows to improve patient care, reduce costs and enhance efficiency. Providers, payers and other stakeholders have also realized many other advantages, including more personalized treatment plans, improved communications across stakeholders and digital transformation across the enterprise.

AI is here to stay in healthcare, and its reach will likely increase. Innovations in AI, such as generative AI (GenAI), agentic AI and intelligent automation, are making waves in the healthcare industry, attempting to solve some of healthcare’s most significant pain points.

This list details, in alphabetical order, the top 12 ways AI has and will continue to impact healthcare.

1. Clinical decision support At its core, a clinical decision support system (CDSS) is a critical tool designed to improve care quality and patient safety. But technologies such as AI and machine learning are transforming clinical decision-making. In the early days of CDSS tools, many were standalone offerings that were not well integrated into clinical workflows. Today, many CDSSes are integrated into electronic health records (EHRs) to help improve deployment and gain more value from the use of these tools at the bedside. This article is part of AI in healthcare: A guide to improving patient care with AI Which also includes:

Health IT (health information technology)

Health IT (health information technology) precision medicine (PM)

precision medicine (PM) telehealth (telemedicine) AI takes this one step further by enabling providers to take advantage of information within the EHR and data pulled from outside of it. Because AI tools can process larger amounts of data more efficiently than other tools while enabling stakeholders to pull fine-grained insights, they have significant potential to transform clinical decision-making. Using AI’s advanced pattern recognition capabilities, CDSS tools can incorporate risk stratification and predictive analytics to help clinicians make more informed, personalized treatment recommendations in high-value use cases, such as chronic disease management.

2. Drug discovery and development Drug discovery, development and manufacturing have created new treatment options for various health conditions. Integrating AI and other technologies into these processes will continue revolutionizing the pharmaceutical industry. High drug development costs and other challenges are driving clinical researchers to seek out new tools to get new drugs to market more efficiently. The process is often high risk, high reward: The drug development lifecycle takes billions of dollars and decades of research, but new medicines aren’t guaranteed to receive regulatory approval from the U.S. Food and Drug Administration. AI and other technologies can help overcome major drug discovery and development barriers. AI and machine learning, in particular, are revolutionizing drug manufacturing by enhancing process optimization, predictive maintenance and quality control while flagging data patterns a human might miss, improving efficiency. These tools are also helpful in the data-gathering systems for complex drug manufacturing, and models to identify novel drug targets are reducing the time and resource investment required for drug discovery. According to pharmaceutical developer Roche, AI is instrumental in creating a more feasible and sustainable timeline for drug development. “The sheer scale and complexity of the scientific data involved in drug discovery pose significant barriers to progress,” the company wrote in a Jan. 30, 2025, post. “Computational approaches have enhanced data collection and analysis, but have historically not matched the magnitude of this problem. Thus, there’s still potential for further advancements in the faster delivery of new medicines and improved success rates in research.” Investments in AI are paying off for pharmaceutical companies. A study published in the June 2024 issue of Drug Discovery Today revealed that AI discovered molecules far better than historic industry averages. The study’s authors suggested that scientists should continue to measure their success as AI-discovered molecules continue down the development pipeline.

3. Electronic health records EHRs hold vast information about a patient’s health and well-being in structured and unstructured formats. This data is valuable for clinicians — but making it accessible and actionable has challenged health systems. AI has given healthcare organizations a unique opportunity to overcome some of these hurdles, and some already see the benefits. EHR adoption aims to streamline clinical workflows while bolstering cost-effective care delivery. However, clinicians cite clinical documentation and administrative tasks as EHR burdens and sources of burnout. AI tools are key to addressing these issues and giving providers back their time so they can focus on patients. There are multiple AI use cases to tackle clinician burnout, most of which aim to automate aspects of the EHR workflow. Health data extraction products can help clinicians find the information they’re looking for quickly and effectively, reducing information overload. Many of these tools use natural language processing (NLP). This AI approach enables algorithms to flag key components of human language and use those insights to parse through text data to extract meaning. AI is also beneficial when healthcare organizations move to new EHR platforms and must undertake legacy data conversion. This process often reveals that patient records are missing, incomplete or inconsistent, which can create significant inefficiencies. Typically, inconsistencies pulled from a medical record require data translation to convert the information into the language of the EHR. The process usually requires humans to translate the data manually, which is time-consuming and labor-intensive and can also introduce new errors that could threaten patient safety. AI-based tools can automate this process, saving time and effort for care teams. Finally, ambient documentation systems powered by AI are instrumental in streamlining provider documentation burdens. Using NLP and machine learning, these tools “listen” to patient-provider conversations during the clinical encounter, transcribe them and then generate a clinical note filed into the EHR for provider review.

4. Genomics Genomics has sparked a wealth of excitement across the healthcare and life sciences industries. Genetic data lets researchers and clinicians better understand what drives patient outcomes, potentially improving care. Particularly, genomics plays a key role in precision and personalized medicine, but making these insights useful requires analyzing large, complex data sets. By enabling providers to combine the power of genomics and big data analytics, AI models can tailor care and treatment recommendations for various medical conditions. These tools are invaluable for overcoming a significant obstacle to using genomics in clinical settings: the data’s actionability. Access to a patient’s genome sequence data sounds promising, as genetic information is relevant to identifying potential health concerns, such as hereditary disease. However, to truly transform care delivery, providers need to know more than just what the data says about a patient’s genetic makeup. They must also determine how that information can be used in the real world. One approach to achieving this involves integrating genomic data into EHRs, which can help providers access and evaluate a more complete picture of a patient’s health. But AI can take this further. “Artificial Intelligence (AI) is valuable in genomics because it enables researchers to analyse vast amounts of complex genomic data more efficiently and accurately than before,” according to an Oct. 17, 2024, blog post by Katrina Costa, a science writer at the Wellcome Sanger Institute. “For example, each human genome contains around 3 billion base pairs and large-scale studies can involve hundreds of thousands of genomes. AI can also help identify patterns and correlations in data that are too subtle or complex for us to detect, and predict the impact of specific changes.” A study published in the May 20, 2024, issue of Nature Communications detailed how an AI-driven model used genomics and epigenetics to assess risk for certain autoimmune diseases. To flag genetic mutations causing certain illnesses, medical researchers must distinguish between cell types — something that’s not always possible. Using an AI-powered tool and genomics, the researchers could predict disease more accurately and thus intervene sooner.

5. Hospital management Managing health system operations is at the heart of how healthcare is delivered. Optimizing workflows and monitoring capacity can have major implications for a healthcare organization’s bottom line and its ability to provide high-quality care. However, monitoring and managing all the resources required is no small undertaking, and health systems are increasingly looking to data analytics tools such as AI to help. Capacity management is a significant challenge for health systems, as issues like ongoing staffing shortages and recent surges in respiratory viruses can exacerbate existing hospital management challenges. Many hospitals, such as Cleveland Clinic, have implemented smart scheduling that uses AI to analyze historical data — including patient volume trends and staff availability — to optimize shift rosters. This type of scheduling can also predict when more staff might be needed, such as during peak flu season and holidays. AI-enabled capacity management is beneficial for surgical scheduling. Since operating rooms are high-cost, high-demand hospital areas, AI can minimize OR downtime by optimizing procedure scheduling and staff availability. Some hospitals have also started using digital twins to improve operational management and performance. Digital twins are virtual replicas of a hospital, including its typical patients, workflows and departments. The technology mirrors data from the EHR, real-time solutions and other IT systems to provide hospital leaders with a platform to test changes and how they might affect care delivery. According to a March 22, 2024, article in npj Digital Medicine, typical applications in healthcare include hospital management, facility design, workflow development, decision-making and individualized therapy. Use of AI in healthcare ranges from chatbots that provide patients with basic information to robot-assisted surgery.

6. Medical imaging Medical imaging is critical in diagnostics and pathology, but effectively interpreting these images requires significant clinical expertise and experience. Imaging analytics, often driven by AI, aims to tackle this. AI technologies are already changing medical imaging by enhancing screening, risk assessment and precision medicine. In a study published in the March 11, 2024, issue of Communications Medicine, Johns Hopkins researchers showed that a deep neural network-based automated detection tool could assist emergency room clinicians in diagnosing COVID-19 by analyzing lung ultrasound images. The tool is designed to identify B-lines — bright, vertical image abnormalities that indicate inflammation in patients with pulmonary complications — with a high degree of accuracy to diagnose COVID-19 infection. The model’s success suggests that a similar approach could be applied to other severe conditions, such as heart failure, to diagnose patients efficiently at the point of care. The researchers emphasized that such a capability would be instrumental in scenarios where emergency department clinicians face high caseloads, like during flu and COVID seasons, or for integration into wearable technologies and other wireless devices for enhanced remote patient monitoring. AI can improve every aspect of a radiologist’s workflow, a top priority for healthcare organizations as the demand for radiologists is expected to grow by almost 26% between 2023 and 2055, according to a study from the Harvey L. Neiman Health Policy Institute published in the February 2025 issue of the Journal of the American College of Radiology.

7. Medical research and clinical trials Medical research is a cornerstone of the healthcare industry, facilitating the development of game-changing treatments and therapies. But this research, particularly clinical trials, requires vast amounts of money, time and resources. AI tools can help researchers overcome the top challenges of clinical trials, including the time it takes to recruit or match patients to a trial, collect large amounts of data from various sources and manually analyze data. AI-powered chatbots can be especially useful for clinical trials to guide patients through eligibility screening and onboarding. These technologies are particularly valuable for accelerating clinical trials by improving trial design, optimizing eligibility screening and enhancing recruitment workflows. Further, AI models can help advance clinical trial data analysis, as they enable researchers to process extensive data sets, detect patterns, predict results and propose treatment strategies informed by patient data. AI has also proven helpful for trial design, enabling protocol simulation to reduce costly amendments.

8. Patient engagement Patient engagement significantly improves health outcomes by enabling patients and their loved ones to be actively involved in care. Patient engagement solutions are often designed to balance convenience and high-quality interpersonal interaction. While digital technologies cannot replace the human elements of the patient experience, they have their place in healthcare consumerism. AI, specifically, can be valuable for personalizing patient engagement tools. Communication is a key aspect of patient experience and activation. EHRs can help facilitate that communication by allowing patients and providers to send messages to one another using the patient portal. However, overflowing inboxes can contribute to clinician burnout, and some queries can be too complex or time-consuming to address using an EHR message. This creates frustration on both sides, as clinicians want to spend more time on care and less on administrative tasks, while patients want their healthcare to be accessible and frictionless. AI chatbots are emerging as a potential solution to this conundrum. They are well-suited to analyzing patient needs and providing resources in certain areas. For example, GenAI is being embedded into more patient portals to address the following two tasks: Mining patient messages and triaging them to the appropriate clinical team member.

Fielding patient messages, analyzing them and generating a response. Studies have shown that GenAI tools provide good medical advice in patient portal messages, although their responses still require review by a healthcare provider. What’s more, chatbots can help filter patient phone calls, sifting out those that can be resolved by providing basic information, such as giving parking information to hospital visitors. The emergence of agentic AI takes this a step further by helping to complete these administrative tasks. These AI tools can also be applied to clinical needs, using patient symptom data to provide care recommendations. AI-driven patient engagement can also take the form of tools designed to conduct patient outreach based on clinical risk assessment data or systems to translate health information for users in a patient portal.

9. Predictive analytics and risk stratification In recent years, the rise of predictive analytics has aided providers in delivering more proactive healthcare to patients. In the era of value-based care, the capability to forecast outcomes is invaluable for developing crucial interventions and guiding clinical decision-making. To successfully use predictive analytics, stakeholders must be able to process vast amounts of high-quality data from multiple sources. For this reason, many predictive modeling tools incorporate AI in some way, and AI-driven predictive analytics technologies have various benefits and high-value use cases. Predictive analytics enables improved clinical decision support, population health management and value-based care delivery, and its healthcare applications are continually expanding. AI-based risk stratification is a crucial component of many of these efforts, as flagging patients at risk for adverse outcomes and preventing those outcomes is integral to advancing high-quality care delivery. For example, researchers at the University Medical Center Groningen developed an AI-driven model to stratify the risk of coronary artery disease (CAD). The model uses an AI-powered questionnaire to forecast a person’s 10-year CAD risk by analyzing their answers about lifestyle, medical history and social factors. The model proved to be as accurate as traditional clinical risk tools that require laboratory analyses, reducing unnecessary utilization. Researchers also touted the cost-effectiveness and scalability of the AI-powered questionnaire, which is more accessible to patients than lab tests. Hospitals are also applying AI capabilities to established predictive analytics solutions that predict adverse events before they happen. Top use cases in this area include risk assessment for sepsis, heart failure and hospital readmissions.

10. Remote patient monitoring Remote patient monitoring (RPM) has become more familiar to patients following the COVID-19 pandemic and the resulting rise in telehealth and virtual care. However, RPM technologies present significant opportunities to enhance patient well-being and improve care by enabling providers and researchers to use additional patient-generated health data. AI can be incorporated into RPM tools or used to streamline RPM data processing. Common RPM tools that use advanced analytics approaches like AI play a significant role in advancing hospital-at-home programs. These initiatives let patients receive care outside the hospital setting, necessitating that clinical decision-making must rely on real-time patient data. RPM offerings enable continuous and intermittent recording and transmission of this data. Tools such as biosensors and wearables are frequently used to help care teams gain insights into a patient’s vital signs or activity levels. AI bolsters these tools’ capabilities by helping to predict complications, helping care teams to preemptively intervene in cases of clinical deterioration, and flagging patients who are likely to benefit from hospital-at-home services compared to inpatient care. These technologies are also helpful because they can learn a patient’s baseline biometrics, detect deviations from that baseline and adjust accordingly or alert the care team when a patient is at high risk for an adverse event.

11. Revenue cycle management Revenue cycle management (RCM) ensures that health systems can focus on providing high-quality patient care. However, effectively tackling revenue challenges and optimizing operations requires heavy lifting on the administrative side. AI tools can help ease these burdens in a variety of ways. RCM still relies heavily on manual processes, but recent trends in AI adoption show that stakeholders are looking at the potential of advanced technologies for automation. Providers are investigating AI-based tools to streamline claims management, which is rife with labor- and resource-intensive tasks, such as managing denials and medical coding. To that end, many in healthcare are interested in AI-enabled autonomous coding, patient estimate automation and prior authorization technology. Healthcare organizations are seeking more information on their ROI before adopting these tools. However, adoption will likely center on operational optimization, leading to automation tools deployed in areas with the highest administrative burden, such as claims management. AI technologies can take over mundane, repetitive tasks — such as checking a claim’s status — and enable staff to focus on more complex revenue cycle management objectives. Revenue cycle management has also been a top target for GenAI in healthcare, considering the relatively low risks of applying the newer technology to administrative versus clinical tasks. For example, GenAI can be used for appointment reminders, preauthorization updates, payment reminders and insurance claim updates. GenAI has also been tapped to improve medical coding by validating codes based on clinical documentation and EHR data and using natural language to turn unstructured data into structured, billing-ready information. Similarly, providers are starting to use GenAI to draft appeal letters for claim denials management, with some AI tools now able to customize denial workflows by payer.

Disclaimer: Information provided by the AMA contained within this Quick Guide is for medical coding guidance purposes only. It does not (i) supersede or replace the AMA’s Current Procedural Terminology (CPT®) manual (“CPT Manual”) or other coding authority, (ii) constitute clinical advice, (iii) address or dictate payer coverage or reimbursement policy, and (iv) substitute for the professional judgement of the practitioner performing a procedure, who remains responsible for correct coding.

CPT © Copyright 2023 American Medical Association. All rights reserved. AMA and CPT are registered trademarks of the American Medical Association. More information can be found on our CPT site.

The policy and payment landscape around telehealth and telemedicine remains complex, and as the public health emergency expires, there will be some uncertainty to navigate related to telehealth policy, coding and payment. The AMA’s Advocacy team will continue to summarize the latest updates in federal and state policy.

The AMA is advocating for you See our real-world impact on issues critical to patients and physicians. Read Impact Report (PDF)

What to expect post PHE What to expect post PHE

The following outlines key policies or actions taken during the COVID-19 pandemic that have been extended past the end of the COVID-19 Public Health Emergency (PHE) either by CMS or through the Consolidated Appropriations Act (CAA) of 2023:

Category 3 telehealth services will be covered through 2023.

Nonfacility payment rates for telehealth services will remain the same through 2023 (physician offices are defined by Medicare as “nonfacility” setting, so this means telehealth payments will remain the same as in-person through 2023.

Direct supervision may continue to be provided virtually through 2023.

Geographic and originating site restrictions on Medicare telehealth services are waived beginning the day of the Public Health Emergency through December 31, 2024.

The in-person requirement on Medicare telemental health services is delayed until on or after January 1, 2025.

Medicare coverage and payment of audio only services will continue through December 31, 2024.

The acute hospital care at home model is extended through 2024.RPM can permanently be used for both chronic and acute conditions.

The ability of Opioid Treatment Programs (OTPs) to provide patient counseling and therapy by phone is permanent.

Recertification of eligibility for hospice care can continue to be conducted via telehealth.

Rural ERs can be originating sites.

An in-person visit will not be required for a patient to be eligible for behavioral health services via telehealth through December 31, 2024.

Codes for reimbursement for audio-only telebehavioral health services which will only be covered through December 31, 2024.

There are also several policy changes and updates to be aware of moving forward including:

CMS decided to continue paying for all of the codes on the telehealth list that were scheduled to stop 151 days after the PHE through the end of 2023, with future policy to be determined in future rulemaking, likely the 2024 Medicare physician payment regulation.

The Drug Enforcement Administration extended its PHE policies on prescribing controlled substances based on telehealth visits for six months after the PHE end until Nov. 11, 2023, to provide time to develop new regulations.

How and when OCR will start to enforce and imposed penalties on covered health care providers for noncompliance with the requirements of the HIPAA Rules in connection with the provision of telehealth using non-public facing audio or video remote communication technologies. The Office for Civil Rights (OCR) is providing a 90-day transition period (PDF) for physicians to come into compliance with the HIPAA Rules regarding telehealth. OCR is responsible for enforcing certain regulations to protect the privacy and security of protected health information, collectively known as the HIPAA Rules. At the beginning of the COVID-19 public health emergency (PHE), the AMA urged policymakers to make it easier for physicians to utilize telehealth in their practice. This included using remote communication technologies for virtual office visits. HIPAA requires that physicians meet certain privacy and security requirements when using remote communication technologies. Responding to AMA advocacy, OCR exercised enforcement discretion (PDF) and did not enforce penalties for physicians who could not or did not meet certain HIPAA requirements. On May 11, the COVID-19 PHE will end and sunset several PHE-related federal policies. The AMA has advocated for additional time for physicians to come into compliance with HIPAA. In response to AMA advocacy, OCR is providing a transition period beginning on May 12 through Aug. 9. OCR recognizes that many physicians began using remote communication technologies for telehealth for the first time during the COVID-19 PHE and need additional time to come into compliance. Therefore, physicians may use this transition period, as necessary, to adjust their telehealth practices to come into compliance, such as by choosing a telehealth technology vendor that will enter into a business associate agreement and comply with applicable requirements of the HIPAA Rules. OCR will continue to exercise its enforcement discretion and will not impose penalties on physicians for noncompliance with the HIPAA Rules during the 90-calendar day transition period.

RPM services were able to be provided to new and established patients during PHE, but following the end of the PHE, there must be an established relationship.

During PHE, RPM could be reported for as few as two days for COVID patients, but this reverts back to 16 days after PHE.

CMS has not been enforcing LCDs on therapeutic continuous glucose monitors during the PHE to allow people with COVID and diabetes to monitor glucose and adjust insulin at home. This waiver ends with the PHE.

CMS decided to not enforce frequency limits for the rest of 2023 and to address them in future rulemaking.

CMS suspended plans to require physicians who provide services from their home to report their home address on their Medicare enrollment through the end of 2023 and will address in future rulemaking.

CMS has agreed to allow MIPS-eligible physicians to request hardship waivers for performance year 2023.

Telehealth services provided by Rural Health Clinics (RHCs) and Federally Qualified Healthcare Centers (FQHCs) can be distant site providers for behavioral health and non-behavioral telehealth services.

Select non-behavioral telehealth services can be provided via audio-only options.

Laws about physician licensure will continue to defer to state law―no federal policy.

In response to AMA advocacy, CMS agreed to continue to allow teaching physicians to provide virtual supervision of residents through 2023 and will address future policy in future rulemaking.

See the full list of authorized telehealth services in the Calendar Year 2023 Medicare Physician Fee Schedule.

Half the dues, all the AMA benefits! Thousands of free CME opportunities to fulfill state requirements

A powerful voice fighting for you during uncertain times

Research, resources, events and more from the largest physician organization Join for Half Dues

Stay up to date on AI Follow the latest news on AI and its applications and effects for health care—delivered to your inbox. Subscribe Now

Policy and payment considerations Policy and payment considerations

Here are some additional key policy and payment considerations to keep in mind:

Ensure that you provide services per your state laws and regulations. As part of emergency declarations, many governors relaxed state laws and regulations related to the provision of telehealth services during the pandemic. States varied on whether these changes were tied to the state PHE or federal PHE, as some of these flexibilities are no longer available. For up to date information in your state, please contact your state department of health or state medical association.

Licensure: There are no additional requirements if you are licensed in the state where the patient is located. If you are not licensed in the state where the patient is located there are several options for physicians to obtain a license or meet the state’s licensure requirements. These options vary by state: Interstate Medical Licensure Compact Licensure by endorsement Special purpose telehealth registry or license Exceptions to in-state licensure requirements CMS has issued the following waiver for Medicare patients (PDF): Temporarily waive requirements that out-of-state clinicians be licensed in the state where they are providing services when they are licensed in another state. Physicians are still bound by their state licensing requirements (CMS FAQs [PDF]). The Federation of State Medical Boards (FSMB) is tracking executive orders related to licensure. Stay up to date on the FSMB website. The AMA has created the following issue brief: Telehealth licensure: Emerging state models of physician licensure flexibility for telehealth (PDF).

CMS expanded access to telemedicine services for all Medicare beneficiaries during the COVID-19 Public Health Emergency. CMS will continue to allow the use of telehealth services until December 31, 2024. In addition to existing coverage for originating sites including physician offices, skilled nursing facilities and hospitals, Medicare will now pay for telehealth services furnished in any healthcare facility and in the home.

AMA provides a resource of COVID-19 flexibilities (PDF) that will end with the PHE.

Telemedicine CPT codes Telemedicine CPT codes

Common CPT codes for telemedicine services are listed below.

The tables on this page give common CPT codes for telemedicine services; other codes may be needed.

Going into 2025, AI is becoming an ever-greater part of our lives. AI is finding its way into a variety of industries, serving B2B interests on the back end and B2C interests on the front end. Sectors ranging from healthcare and finance to manufacturing, retail and education are automating routine tasks, improving UX and enhancing decision-making processes with the technology.

AI is also moving out of the data center and into the world through smartphones, IoT devices, autonomous cars and other intelligent instruments that interact with their environments. Improvements in real-time processing, lower latency, enhanced privacy and reduced bandwidth usage will make these embodied AI machines more efficient and safer.

At the same time, there remains a strong focus on the ethical use of AI, with an emphasis on fairness, transparency, explainability and accountability in AI models and decision-making processes. This is a departure from most technological advances, where ethics often play catch-up after adoption takes off.

All this AI growth means more jobs. Below is a discussion of the skills companies are looking for in an AI specialist, the industries that are aggressively adopting AI and a list of what might be the 10 hottest AI jobs and skills for 2025.

Top AI job skills Many programmers across all fields are self-taught, and the resources available online make it easy for novices to educate themselves on popular languages, like C++, Java and Python. But the AI space has much higher demands. National University examined 15,000 job postings on Indeed to determine the requirements for AI jobs. It found 77% of AI job openings required that candidates have a master’s degree, outpacing the 69% of postings that required at least a bachelor’s degree. Another 18% required a doctoral degree, while only 8% of jobs posted were open to candidates with just a high school diploma. The job openings predominantly required a moderate amount of experience, with midlevel positions accounting for almost half the job openings (44%), followed by senior-level (26%) and entry-level (12%) roles. There were no jobs that called for no prior experience. And, while many types of IT jobs are remote work-friendly, AI jobs are not. Only 11% of job openings offered fully remote work, and another 15% allowed for a hybrid situation of on-premises work and remote work. The remaining 74% required on-site presence. In short, being a successful AI developer requires more than just coding skills. Proficiency in a core AI developer language, such as Python, Java or R, along with emerging languages, such as Julia or Scala, is essential. That alone won’t land you a job, however. According to ZipRecruiter, programming is just one of the following five top required skills for AI programming jobs: Communication skills.

Knowledge and experience with Python specifically.

Digital marketing goals and strategies.

Effective collaboration with others.

Analytical skills. AI jobs demand critical thinking skills on the part of developers to solve problems and analyze user input. The same practices apply to code. Having strong mathematical skills can help people develop advanced algorithms for programs. AI is also unique because it requires some knowledge of psychology because AI simulates human behavior. To create AI, people need to understand how humans think and how they might behave in different situations. Finally, with an emphasis on AI security, privacy and data integrity, individuals need to know the best practices behind security and ethics. Some industries are embracing AI faster than others. These include the following: Technology. Tech firms of all types are adding AI to their products to enhance their use and make them simpler and more user-friendly. Hyperscalers, such as Google, Amazon and Microsoft, are all actively hiring AI specialists to build services.

Tech firms of all types are adding AI to their products to enhance their use and make them simpler and more user-friendly. Hyperscalers, such as Google, Amazon and Microsoft, are all actively hiring AI specialists to build services. Finance. The finance industry is making broad use of AI with simple tasks, such as automation, and more advanced uses, including improving risk management and making better investment recommendations and decisions.

The finance industry is making broad use of AI with simple tasks, such as automation, and more advanced uses, including improving risk management and making better investment recommendations and decisions. Healthcare. The healthcare industry is also rapidly embracing AI at all levels. On the low end, AI is being used for automation to avoid human error and for tasks such as billing and record management. On the high end, AI is being widely touted for early detection of serious illnesses, such as cancer, because AI can spot signs that humans might miss.

The healthcare industry is also rapidly embracing AI at all levels. On the low end, AI is being used for automation to avoid human error and for tasks such as billing and record management. On the high end, AI is being widely touted for early detection of serious illnesses, such as cancer, because AI can spot signs that humans might miss. Retail. The retail industry is making wide use of AI for operational efficiency. AI can be used for areas such as inventory management, loss prevention, trend spotting, more personal shopping experiences and fraud prevention by finding suspicious spending patterns or transactions.

The retail industry is making wide use of AI for operational efficiency. AI can be used for areas such as inventory management, loss prevention, trend spotting, more personal shopping experiences and fraud prevention by finding suspicious spending patterns or transactions. Manufacturing. The manufacturing industry is embracing AI for operational efficiency. AI can provide early detection of potential equipment failure and help machinery run efficiently.

The manufacturing industry is embracing AI for operational efficiency. AI can provide early detection of potential equipment failure and help machinery run efficiently. Cybersecurity. The cybersecurity market is embracing AI to monitor threats around the clock and to avoid human error. AI applications can be programmed to detect unusual activity quickly for swift action.

Background The rapid expansion of artificial intelligence (AI) in 2023 propelled it into the spotlight, prompting increasing questions over how AI could be used to support the healthcare system. Deployed well and in the appropriate contexts, AI has the potential to help the healthcare system overcome a range of challenges. With the field of AI evolving at pace, this primer has been developed to support healthcare organisations to navigate the noise. Intended for board members and wider teams, it demystifies the language of AI and showcases how organisations and systems are using it – and to what end. With more and more organisations exploring AI’s potential, this primer also suggests how you can use early conversations with suppliers to consider whether an AI solution is what you need.

Demystifying AI in healthcare: the jargon buster Please note that not all definitions have an associated example as in many cases an example would be repeating the definition.

The basics

Automation/robotic process automation (RPA) Plus icon The use of specialised software and technology to carry out repetitive tasks, following a set of instructions and workflows set out by humans. These tasks typically remain consistent over time and include actions like sending appointment reminders, missed appointment notifications, or even receipts after online purchases. If a task is not explicitly outlined in the instructions, the machine cannot perform it. Example : In healthcare, automation extends to patient monitoring, medication management and administrative tasks in hospitals and clinics.

Algorithm Plus icon A set of well-defined rules or processes used by an AI system to conduct tasks such as discovering new insights, identifying patterns, predicting outcomes and solving problems.

Artificial intelligence (AI) Plus icon The capability of a computer system to mimic human cognitive functions such as learning, problem-solving, interpreting visual information, understanding, and responding to spoken or written language. AI uses maths, logic and patterns learned from data to simulate human reasoning and make decisions and recommendations. Example : In healthcare, AI can be used to enhance diagnostic processes, personalise treatment plans and manage healthcare data efficiently.

Data Plus icon Any information that can be processed or analysed to gain insights. Data can take the form of numbers and statistics, text, symbols, or multimedia such as images, videos, sounds and maps. Example : In the context of healthcare, data can encompass patient records, clinical studies and real-time health monitoring outputs.

Machine learning (ML) Plus icon A subset of AI that enables machines to automatically learn and improve from experience without explicit programming. By using set processes to analyse large amounts of data, ML systems can identify patterns, help make decisions, and improve their performance with little to no human intervention. Example : In healthcare, ML applications include predicting disease progression, analysing medical images and optimising clinical workflows.

Model Plus icon A simple representation of an aspect of the real world. It is a programme that has been trained on a set of data to recognise certain patterns or make certain decisions without further human intervention.

Prompt (engineering) Plus icon A prompt is a question, command or statement input into an AI model to initiate a response or action, facilitating interaction between a human and the AI to generate the intended output.

Interpreting models

Accuracy Plus icon Accuracy is a metric in machine learning that measures how often the model correctly predicts the outcome. It is the fraction of predictions that the model got right, indicating the overall correctness of the model’s predictions. It shows how often a classification ML model is correct overall. Accuracy is useful when the classes are balanced (ie the number of instances in each class is roughly the same). However, it can be misleading in cases of imbalanced classes. Ideally, accuracy should be as close to 100 per cent as possible, however, 70-90 per cent are often cited as acceptable ranges. It is important to remember that 50 per cent accuracy means 50 per cent are classified as positive and 50 per cent as negative, which is essentially the same as random classification.

Bias Plus icon Bias occurs when an AI system produces results that are systematically prejudiced due to flawed assumptions in the machine learning process. This bias can reflect and perpetuate human biases and social inequalities present in the initial training data, the algorithm itself, or the predictions it generates. Example : Pulse oximeters are less accurate for people with darker skin tones, meaning AI applied to this device can underestimate skin cancer in people with darker skin due to less data.

Explainability Plus icon A measure of how understandable, or explainable, the decisions of an AI system are to humans. Example : An AI system may predict which patients are most in need of surgery but should be able to explain why it has prioritised patients in a certain way.

Explainable AI (XAI) Plus icon Where humans can understand how the results of an AI model were obtained.

Model drift Plus icon This is the degradation of a machine learning model’s predictive accuracy over time, caused by changes in real-world environments or new input data differing from the data used during training. Example : When a new bus route opens to a hospital making a model used to predict did-not -attends less accurate at predicting attendance patterns for patients, due to different trends compared to when the model was trained (prior to the new bus route).

Precision Plus icon Precision is the proportion of positive class predictions that were actually correct. For instance, if the model predicts 100 instances as positive and 70 of them are truly positive, the precision is 70 per cent. Precision shows how often an ML model is correct when predicting the target class.

Scalability Plus icon ML scalability refers to the capability of a machine learning system to handle increasing amounts of data and computational resources without compromising performance or precision. It involves the ability to process large datasets while still producing accurate results in a reasonable amount of time.

Sensitivity (recall) Plus icon Sensitivity is the proportion of actual positive class instances that the model correctly identified. For example, if a dataset has 100 positive instances and the model correctly identifies 60 of them, the recall is 60 per cent. Recall measures the ability of the machine learning model to identify all objects of the target class.

Specificity Plus icon Specificity indicates the model’s ability to accurately predict true negatives for each category. In other words, specificity assesses how well the model correctly identifies instances that do not belong to the target class.

Training a model Plus icon Training a model in machine learning is the process of teaching a machine learning algorithm to make predictions or decisions based on data. Example: In healthcare, this often involves training with clinical data to improve accuracy in diagnosis and treatment efficacy.

Types of data

Big data Plus icon Extremely large and rapidly growing collections of diverse data types including, structured and unstructured, which are so complex that traditional data processing software cannot handle them. Example : In healthcare, big data can include processing multiple structured and unstructured data sources, such as genetic data, medical history, and lifestyle factors to support personalised medicine.

Structured data Plus icon Data that is organised and formatted in a specific way, making it easily readable and understandable by both humans and machines, allowing viewers to immediately recognise the type of data they are looking at. Example : A patient’s electronic health record (EHR) that includes fields for name, age, blood pressure and diagnosis codes is structured data.

Synthetic data Plus icon This is artificially generated data produced by computer algorithms or simulations, designed to mimic the patterns and characteristics of real-world data, and often used as an alternative to actual data.

Test data Plus icon A final check of an unseen dataset to confirm that the ML algorithm was trained effectively and validate that the model can make accurate predictions.

Training data Plus icon The data used to train machine learning models. Curated training datasets are fed to machine learning algorithms to teach them how to make predictions or perform a desired task.

Unstructured data Plus icon Data that does not have predefined structure or organisation. Unlike structured data, which is organised into neat rows and columns in a database, unstructured data is an unsorted and vast information collection. Example : In healthcare, unstructured data often includes medical notes, audio recordings of patient interactions and images from various diagnostic procedures.

Validation data Plus icon Data not included in the training set of the model, allowing data scientists to evaluate how well (using metrics like accuracy, precision, sensitivity and specificity) the model makes predictions based on new data unseen by the model as it is being trained.

Types of machine learning

(Artificial) neural network Plus icon A neural network is a type of machine learning programme that makes decisions similarly to the human brain. It processes data using interconnected units called neurons, which work together to identify patterns, weigh options, arrive at conclusions and learn and improve over time. This method, inspired by how biological neurons function, teaches computers to handle complex problems by mimicking the brain’s layered structure.

Reinforcement machine learning Plus icon A subset of machine learning that allows an AI-driven system to learn through trial and error, using feedback from its actions. Example : This is particularly useful in personalized medicine, where systems learn to optimise treatments based on individual patient responses.

Semi-supervised machine learning Plus icon A type of machine learning that falls in between supervised and unsupervised learning. It is a method that uses a small amount of labelled data and a large amount of unlabelled data to train a model. Example : This approach is beneficial for patient data where obtaining fully-labelled datasets can be costly or impractical.

Supervised machine learning Plus icon A category of machine learning where labelled datasets (each input has a known output) are used to train algorithms to predict outcomes or recognise patterns. By studying these datasets, the computer learns to predict the output given new input data. It is like teaching a computer by showing it many examples and letting it figure out how to do things correctly. Example : In healthcare, this method is extensively used for diagnostics, such as identifying diseases from medical imaging data.

Unsupervised machine learning Plus icon A type of machine learning that does not need labelled data or human guidance. It works with unlabelled data to discover patterns and insights within the dataset. The algorithms explore the dataset without explicit instructions to find unknown relationships or insights independently. It is like letting the computer explore the dataset with the teacher allowing it to uncover patterns and structures by itself.

Types of model

Deep learning model Plus icon A form of machine learning that employs artificial neural networks, inspired by the human brain, to learn from vast amounts of data (including labelled and unlabelled, structured and unstructured data). These networks enable the digital systems to learn and make decisions automatically and independently without human intervention. Example : These models are increasingly used in areas such as pathology, radiology, and genomics.

Foundation model Plus icon A machine learning model trained on a vast amount of data so that it can be easily adapted for a wide range of applications. A common type of foundation model is large language models, which power chatbots such as ChatGPT.

Human-in-the-loop Plus icon A system comprising a human and an AI component, in which the human can intervene in some significant way, such as by training, tuning or testing the system’s algorithm, so that it produces more useful results. It is a way of combining human and machine intelligence, helping to make up for the shortcomings of both.

Large language model (LLM) Plus icon A machine learning model capable of performing various natural language processing tasks. These tasks include generating and classifying texts and images, answering questions conversationally, translating between languages, predicting, and summarising content. It uses deep learning algorithms and a vast dataset to achieve these capabilities. Example : In healthcare, these models assist in clinical decision support and patient interaction.

Multimodal model Plus icon This is a machine learning model that processes and combines different types of data, such as images, videos and text, to make more accurate determinations, draw insightful conclusions, or make precise predictions about real-world problems. Example: Multimodal models can include data from an electronic health record, an image captured by an X-ray and a radiologists written description of an X-ray to derive conclusions around diagnoses.

Applications of AI

AI hallucination Plus icon An AI hallucination occurs when an AI, such as a large language model, produces false or misleading information that seems factual but is actually inaccurate or nonsensical. This can be through identifying patterns that do not exist in real life. Example : For example, an AI model suggesting the wrong medication for a patient based on hallucinated data.

Ambient AI Plus icon Ambient AI is a type of AI that blends into the environment to improve human interaction without being noticeable. It works quietly in the background, using sensors to understand and predict human behaviours. It continuously collects data from these devices like sensors to make real-time decisions. Example : In healthcare, ambient AI can be used to monitor patient conditions in real-time, optimise hospital operations and deliver personalised healthcare services, all while minimising the need for direct human command or intervention. It enhances patient care by predicting needs and intervening proactively, thereby improving patient outcomes and operational efficiency.

Computer vision Plus icon A field of AI that trains computers to interpret and understand the visual world. Machines can accurately identify and locate objects and then react to what they ‘see’ using digital images from cameras, videos and deep learning models. Example: Computer vision is used in tools that automatically screen for diabetic retinopathy from retinal images.

Decision support system Plus icon A computer-based system that helps users make decisions by analysing large amounts of data, providing insights and suggesting possible courses of action. It combines data, analytical models and user-friendly software to support problem-solving and decision-making. Example : In healthcare, this could include a machine learning algorithm that analyses radiology images to provide a diagnosis to support physician decision making.

Digital twin Plus icon A computer model that simulates an object in the real world, such as a biological system. Analysing the model’s output can tell researchers how the physical object will behave, helping them to improve its real-world design and/or functioning.

Generative AI (Gen AI) Plus icon Algorithms capable of creating new original content, including text, images, audio, simulations and software code, in response to user prompts or requests. Example : In the medical field, generative AI is used to simulate patient data, develop virtual models for training, and generate synthetic biological data for research.

Natural language processing (NLP) Plus icon Predictive analytics is the process of using data to forecast future outcomes. The process uses data analysis, machine learning, AI, and statistical models to find patterns that might predict future behaviour. Example : Healthcare applications include predicting disease outbreaks, patient deterioration, and therapy outcomes.

Predictive analytics/predictive modelling Plus icon Predictive analytics is the process of using data to forecast future outcomes. The process uses data analysis, machine learning, AI , and statistical models to find patterns that might predict future behaviour.

How the healthcare system is using AI The potential uses of AI are widespread, from clinical uses to administrative support, in primary care settings and hospital settings. Below we have highlighted examples from within our membership that showcase the breadth of potential AI use within the healthcare system. These examples also put the AI jargon into a real-life context to help further understanding of key terminology.

Reducing DNAs and last-minute cancellations

Background Mid and South Essex (MSE) NHS Foundation Trust supports a population of 1.2 million people. It had an average did-not-attend (DNA) and short-notice cancellation rate of 8 per cent (the average in England is approximately 7.6 per cent). DNAs are more prominent in patients grappling with work and caring responsibilities, who find it difficult travelling to the hospital at a specific time. This disproportionately affects patients from minority and marginalised communities and can thereby lead to a widening of health inequalities. MSE found that its DNA rate for the most deprived was consistently higher than that for the least deprived across all age groups. Previously Many practices and hospitals send letters, emails, text and calls to remind patients of their appointments. These blanket approaches do not account for individual patient behaviours. Healthcare staff may manually review appointment histories and contact patients who had missed appointments in the past, but this is time and resource intensive. Some organisations may overbook certain clinics to compensate for anticipated no shows. Solution MSE rolled out Deep Medical, which uses a model that integrates machine learning to predict patient no-shows and short-notice cancellations (<48 hours). Using AI, the system processes both structured data (like patient demographics and appointment history) and unstructured data (like electronic health record notes) to identify relevant patterns. The predictive analytics model is developed through a deep learning model which is a subset of supervised machine learning. This model is trained using training data, fine-tuned with validation data and evaluated with the test data to ensure the model is accurate and precise and has high sensitivity and specificity. The neural network format of this model allows it to recognise complex patterns in real-world data. It uses AI to understand patient engagement and develop personalised reminder schedules for patients. It then uses automation to send these reminders and manage patient appointments. It also identifies frail patients where low compliance could potentially be of clinical concern and highlights them to the relevant clinical teams. Outcomes Based on a six-month pilot scheme at MSE, the model: created an additional 1,910 patient visits into clinic

prevented 377 DNAs

filled 217 last-minute cancellation spots ️

reduced DNA rate by up to 50 per cent when patients are proactively contacted two to three weeks ahead of their appointments

when used at full scale, it is predicted it will allow an additional 80,000 patients to be seen each year at the trust, increasing productivity significantly Top tips Make sure you are clear on: what the specificity and sensitivity of the data is and what the impact of the mis-classified data is

where the training data has come from and how valid it is in your local population or whether revalidation is required

how to audit the machine learning to ensure it doesn’t drift from the primary use or introduce new biases.

Transforming wound care

Background In North Cumbria approximately 50 per cent of the community nurses' workload involves managing both acute and chronic wounds, which amounts to the cost of around £41.7 million a year. Previously Prior to using the AI model, community nurses would assess wounds with a manual tape measure. Band 3 and 4 staff are sometimes unsure how to dress and treat more complex wounds, therefore, this would require taking a photo and sending it back to base where a band 6 or 7 would tell them how to treat the wound by looking at photos and advising how to change and apply the dressing remotely. In more severe cases the band 6/7 would need to go out to treat the wounds themselves, reducing time to care. This was a very subjective way to treat and dress the wounds with a lot of room for interpretation and no consistency across community teams. There was no way to keep a record of the wounds and the changes. Solution North Cumbria Integrated Care (NCIC) Foundation Trust, as part of North East and North Cumbria (NENC) ICB, transformed its approach to wound care by introducing a digital tool, Minuteful for Wound by Healthy.io, that uses a model which uses an AI algorithm to support clinicians at various experience levels to assess wounds confidently, consistently and safely. This model offers a higher degree of accuracy and precision than the current standard care. The digital tool enables consistent wound imagery, measurements and assessments to be captured easily at the point of care. The AI-powered colour recognition technology automatically detects wound area and tissue types within the wound bed from a three-second video scan of the wound by a smartphone. The AI also ensures image quality by recognising lighting conditions, distance and scan technique. Assessment prompts within the app guide allow clinicians of all experience levels to record assessments aligned to best practice guidance according to wound type. The approach offers a live caseload review portal to provide central access to wound data. With this easily accessible data, clinical teams are able to optimise care plans. The data gathered from AI measurements and assessments enables wounds that may be deteriorating or static over time to be identified and flagged for review that might otherwise not be identified. The new approach includes assigning clinicians to perform regular virtual caseload reviews within the tool's portal. This earlier detection and ability to review the caseload virtually, leads to earlier intervention and faster healing for patients. Outcomes Since using the model, NCIC has made the following positive impact for both staff and patients. For the workforce, it has: opened access to data to help them understand wound care admission and case loads

upskilled and retrained the workforce to dress and treat wounds with confidence, which has led to greater empowerment of junior clinical staff to undertake assessments

provided assurance for staff as they are creating a log of the wound which provides detail on how the wound was treated if the patient is admitted into hospital

released time to care for band 6/7 staff

reduced admin burden for staff. For patients, it has: reduced healing time and improved patient outcomes

reduced hospital admissions

supported self-care

minimised the likelihood of future infections, admissions to hospital and amputations. By working in partnership with the company, NENC ICB is hoping to: reduce incidents by early identification of deteriorating wounds and clinical risks

standardise clinical data collection and reporting

improve consistency of adherence to formulary and best-practice guidelines

increase system-wide collaboration with ICS partners to expand digital wound management beyond the trust, to primary care, acute services and nursing/care home settings

enable access to services, including emergency departments/vascular specialists both via app and the web page portal for greater information sharing. Top tips Make sure you are clear on: what IT infrastructure and resources are required for roll out and whether you have these already

what training, standard operating procedures and ways of working are required for an effective roll out. Impact on the workforce To find out more about the benefits of this project for the workforce as well as patients, please read this case study.

Improving patient triage and optimising staff time

Background Chapelford Medical Centre wanted to update its systems to triage patients in a smarter way, helping them get the care they need while also optimising use of staff time. Previously Chapelford Medical Centre had a triage tool and questionnaire to collate information that supported staff with triaging patients. This used automation and operated like a decision tree. The centre wanted to move to an approach that allowed evidenced-based and AI-supported decisions about the best course of action for a patient’s care. The ultimate goal was to make informed decisions and spot opportunities that would not be possible with a human-only approach, based on combining the information presented by the patient with information within patient records. Solution Anima, an integrated care platform, has both automation and AI modules. In the automation module, patients complete a questionnaire to support with triage and based on their answers, they will be directed to an appropriate care pathway. In the AI module, Anima uses analysis of the structured and unstructured data input into the questionnaire by patients, and machine learning based on the outcomes of patients who go through the pathway to predict the best course of action for future patients. It uses a model and algorithm to follow the pathway of patients through triage and their outcomes, and refine the course of action for future patients with similar characteristics who fill in the questionnaire. Outcomes Anima went live within the GP practice in August 2023 and the ambition is to expand across the primary care network from August 2024. The Chapelford Medical Centre team plans to be in contact with Anima regularly and provide mutual training support. The team will review and work out whether improvements can be made to the AI suggestion, and also improve their own experiential learning for next time by using a trusted source of information. They are aiming to create an enhanced access system based on need, and not ‘first in the queue’. This includes better management of patients, ultimately less demand, an improved workload profile for clinicians, and a fully-functional multi-disciplinary team. AI will support the team at each stage, enabling staff to signpost effectively, and employ more junior team members to make effective and supported decisions. Access increases, patient need decreases, clinician workload levels out, and work-life balance improves. Chapelford hopes to achieve a position where all patients get the care they need every time, first time. This is a challenge, but one that will significantly improve patient journey, and also maximise the potential of appointments by managing patients within one consultation. They aim to maximise individual clinician’s skills and achieve a 0 per cent need for referral between teams for the presenting condition. Numbers of staff are unlikely to increase, so Chapelford hopes to increase effective capacity through more efficient use of the team and getting it right first time. By feeding back through an AI system, the system will continually improve and have a real-time feedback loop. Top tips Make sure you are clear on: which elements of the solution use AI and that it really is AI

what the AI is improving or making better, and how it is measured

what the clinical safety case is for the solution you are buying

what your North Star is and whether the AI product feeds into a bigger and wider solution.

Transforming the cataract care pathway

Background The cataract care pathway at Chelsea and Westminster Hospital was experiencing long waiting times and limited capacity, Previously The pathway required patients to attend a mandatory in-person pre-operative appointment before being placed on the surgery waiting list for their first eye. The did-not-attend (DNA) rate for these pre-operative clinics ranged from 30 per cent to 50 per cent, resulting in wasted clinical capacity. Post surgery, although there are very low risks of complications and adverse outcomes from cataract surgery (less than 5 per cent), all post-operative patients were requested to attend an in-person appointment four weeks later. After consultation, patients were listed for their second eye surgery or discharged to the community. Solution Chelsea and Westminster Foundation Trust leveraged Ufonia’s Dora, an AI clinical assistant that can telephone patients and have a routine clinical conversation using AI-enabled automation, to transform their cataract care pathway. Dora has a telephone-based voice conversation with patients at multiple timepoints in their cataract journey. This includes pre-operative assessment health screening, surgery and appointment reminders, post-operative checks and patient-reported outcome measures. The technology uses AI natural language processing to be able to interpret a patient's responses. It focuses on making AI accessible, meaning patients do not require any technology understanding, user accounts, hardware devices or training – they simply have a conversation. Outcomes Dora was initially deployed as a pilot, with pre-defined technical and operational success criteria. These were met during the initial phase meaning the trust now plans to adopt Dora to support their cataract patient pathways, advocate for regional implementation across North West London ICB, and is considering expanding the solution to other clinical pathways. Successes include: 65 per cent call completion rate at pre-operative assessment stage, surpassing the original target of 60 per cent

91 per cent agreement rate between Dora and Chelsea and Westminster Hospital clinical staff, exceeding the goal of 90 per cent

on-the-day cancellation rate has dropped significantly to just one patient in six months, compared to four patients per month before implementing Dora, surpassing the target of two patients per month.

fewer than 2.5 per cent of patients at the post-operative check experienced an unexpected management change , well below our target of 10 per cent

63 per cent of completed calls passed the Dora assessment , exceeding the target of 50 per cent

honoured as finalists in the HSJ Digital Awards 2024 for the Driving Change through AI and Automation Award. Top tips Make sure you are clear on: what implementation support exists from the tech side to ensure key milestones can be met

what resources might be required and what lessons could be learned from previous implementations, to support internal resource allocation

which metrics are being collected and when, to ensure robust evaluation following implementation.

To AI or not to AI? Key questions to consider It is important to remember that AI may not be the appropriate solution to the problem your organisation is facing. Start with the challenge and ensure you select a solution that solves the issue. To support with innovation adoption, please refer to our Scaling Innovation guide. To help determine whether AI is the right fit for solving the problem, and whether further exploration of the solution is required, we have prepared a set of key questions to ask suppliers of AI solutions in early discussions: Questions to ask suppliers What to look out for What is this AI solution designed to do? Do the intended uses of the AI solve the issues you are trying to solve? What uses does the AI have regulatory approval and certifications for? Are there any exclusions or limitations you need to be aware of? 2. What evidence is there supporting the effectiveness of the AI? Does evidence exist in real-world settings? Has the solution been tested in a real-world clinical setting? How do the results compare to effectiveness without use of the AI component? Is the AI adding value? How similar is the training data cohort to your population? Are there potential biases that have been raised that you need to consider and explore in more depth? Does the supplier indicate if and how they mitigate against any biases? Has another organisation rolled out the product that you could get in contact with to gather their experience? 3. What implementation support within your existing infrastructure is required? What level of effort will be required to integrate the solution into existing digital pathways and for ongoing monitoring? Will any changes be required in IT and digital systems and data management? Has another organisation rolled out the product that you could get in contact with to gather their experience? These questions can be used to gather an understanding of the purpose of the proposed AI, the evidence supporting its use and the implementation support required. If after hearing the responses you feel the product may be a good solution to the problem you are trying to solve, deeper exploration is required before a final procurement decision can be made. Next steps if moving forward with an AI solution Following the initial conversation, if you have decided the product could be a good solution, assemble a team of experts from within your organisation with expertise covering data and IT infrastructure, information governance, implementation and transformation, cyber security and clinical leadership (if the AI has a clinical component) to further explore the feasibility of adopting the AI into your organisation. There are several resources that can support AI adopters with navigating the complex environment of adopting AI within the NHS, including: A Buyer’s Guide to AI in Health and Care , developed by NHS Transformation Directorate in 2020 contains additional questions and detailed answers. The majority of the information is still relevant in the 2024 environment; however, elements of the regulatory environment section may not reflect the current situation.

developed by NHS Transformation Directorate in 2020 contains additional questions and detailed answers. The majority of the information is still relevant in the 2024 environment; however, elements of the regulatory environment section may not reflect the current situation. Developing Healthcare Workers’ Confidence in AI report developed by the NHS AI Lab and Health Education England in 2022 contains additional questions to consider when conversations with suppliers develop (pages 88-91).

All Adopters' Guidance lists of all the guidance and regulations that apply to adopters of digital technologies in health and social care.

Artificial Intelligence where the NHS Digital Academy has curated resources including a framework to support healthcare workers understand their level of understanding in AI.

Adopt AI on the NHS England Transformation Directorate website curates a selection of resources.