Automated medical scribe

Last updated

Automated medical scribes (also called artificial intelligence scribes, AI scribes, digital scribes, virtual scribes, ambient AI scribes, AI documentation assistants, and digital/virtual/smart clinical assistants [1] ) are tools for transcribing medical speech, such as patient consultations and dictated medical notes. Many also produce summaries of consultations. Automated medical scribes based on Large Language Models (LLMs, commonly called "AI", short for "artificial intelligence") increased drastically in popularity in 2024. There are privacy and antitrust concerns. Accuracy concerns also exist, and intensify in situations in which tools try to go beyond transcribing and summarizing, and are asked to format information by its meaning, since LLMs do not deal well with meaning (see weak artificial intelligence). Medics using these scribes are generally expected to understand the ethical and legal considerations, and supervise the outputs.

Contents

The privacy protections of automated medical scribes vary widely. While it is possible to do all the transcription and summarizing locally, with no connection to the internet, most closed-source providers require that data be sent to their own servers over the internet, processed there, and the results sent back (as with digital voice assistants). Some retailers say their tools use zero-knowledge encryption (meaning that the service provider can't access the data). Others explicitly say that they use patient data to train their AIs, or rent or resell it to third parties; the nature of privacy protections used in such situations is unclear, and they are likely not to be fully effective. [2] [1] [3] [4]

Most providers have not published any safety or utility data in academic journals, [1] and are not responsive to requests from medical researchers studying their products. [5]

Privacy

Some providers unclear about what happens to user data. [3] Some may sell data to third parties. [1] Some explicitly send user data to for-profit tech companies for secondary purposes, [1] which may not be specified. Some require users to sign consents to such reuse of their data. [2] Some ingest user data to train the software, [1] promising to anonymize it; however, deanonymization may be possible (that is, it may become obvious who the patient is). [3] It is intrinsically impossible to prevent an LLM from correlating its inputs; they work by finding similar patterns across very large data sets. Some information on the patient will be known from other sources (for instance, information that they were injured in an incident on a certain day might be available from the news media; information that they attended specific appointment locations at specific times is probably available to their cellphone provider/apps/data brokers; information about when they had a baby is probably implied by their online shopping records; and they might mention lifestyle changes to their doctor and on a forum or blog). The software may correlate such information with the "anonymized" clinical consultation record, and, asked about the named patient, provide information which they only told their doctor privately. Because a patient's record is all about the same patient, it is all unavoidably linked; in very many cases, medical histories are intrinsically identifiable. [6] Depending on how common a condition and what other data is available, K-anonymity may be useless. Differential privacy could theoretically preserve privacy.

Data broker companies like Google, Amazon, and Microsoft have produced or bought up medical scribes, [5] some of which use user data for secondary purposes, [2] which has led to antitrust concerns. [7] Transfer of patient records for AI training has, in the past, prompted legal action. [8]

Open-source programs typically do all the transcription locally, on the doctor's own computer. [9] [10] Open-source software is widely used in healthcare, with some national public healthcare bodies holding hack days. [11]

Encryption

Multifactor authentication for access to the data is expected practice. [1]

Typically, Diffie–Hellman key exchange is used for encryption; this is the standard method commonly used for things like online banking. This encryption is expensive but not impossible to break; it is not generally considered safe against eavesdroppers with the resources of a nation-state. [4]

If content is encrypted between the client and the service provider's remote server (transport cryptography), then the server has an unencrypted copy. This is necessary if the data is used by the service provider (for instance, to train the software). Zero-knowledge encryption implies that the only unencrypted copy is at the client, and the server cannot decrypt the data any more easily than a monster-in-the-middle attacker.

Platforms

Scribes may operate on desktops, laptop, or mobile computers, under a variety of operating systems. These vary in their risks; for instance, mobiles can be lost. [12] [13] [14] [15] The underlying mobile or desktop operating systems are also part of the trusted computing base, and if they are not secure, the software relying on them cannot be secure either. [16] [17]

Confabulation, omissions, and other errors

Like other LLMs, medical-scribe LLMs are prone to confabulation, where they make up content based on statistically associations between their training data and the transcription audio. [3] LLMs do not distinguish between trying to transcribe the audio and guessing what words will come next, but perform both processes mixed together. [18] They are especially likely to take short silences or non-speech noises and invent some sort of speech to transcribe them as. [18] [2]

LLM medical scribes have been known to confabulate racist and otherwise prejudiced content; this is partly because the training datasets of many LLMs contain pseudoscientific texts about medical racism. They may misgender patients. [3] A survey found that most doctors preferred, in principle, that scribes be trained on data reviewed by medical subject experts. [19] Relevant, accurate training data increases the probability of an accurate transcription, but does not guarantee accuracy. [18] Software trained on thousands of real clinical conversations generated transcripts with lower word error rates. Software trained on manually-transcribed training data did better than software trained with automatically transcribed training data [5] (such as YouTube captions).

Autoscribes omit parts of the conversation classes as irrelevant. The may wrongly classify pertinent information as irrelevant and omit it. They may also confuse historic and current symptoms, or otherwise misclassify information. They may also simply wrongly transcribe the speech, writing something incorrect instead. If clinicians do not carefully check the recording, such mistakes could make their way into their medical records and cause patient harms. [1]

Professional organizations generally require that scribes be used only with patient consent; some bodies may require written consent. Medics must also abide by local surveillance laws, which may criminalize recording private conversations without consent. [1] Full information on how data is encrypted, transmitted, stored, and destroyed should be provided. In some jurisdictions, it is illegal to transmit the data to any country without equivalent privacy laws, or process or store the data there; vendors who cannot guarantee that their products won't illegally send data abroad cannot be legally used. [1]

Some vendors collect data for reuse or resale. Medical professionals are generally considered to have a duty to review the terms and conditions of the user agreement and identify such data reuse. [1] General practices are generally required to provide information on secondary uses to patients, allow them to opt out of secondary uses, and obtain consent for each specific secondary use. Data must only be used for agreed-upon purposes. [1] [20]

Technology and market

The medical scribe market is, as of 2024, highly competitive, with over 50 products on the market. Many of these products are just proprietary wrappers around the same LLM backends, [7] including backends whose designers have warned they are not to be used for critical applications like medicine. [21] Some vendors market scribes specialized to specific branches of medicine (though most target general practitioners, who make up about a third of doctors[ where? ]). Increasingly, vendors market their products as more than scribes, claiming that they are intelligent assistants and co-pilots to doctors. [7] These broader uses raise more accuracy concerns. [18] [2] Extracting information from the conversation to autopopulate a form, for instance, may be problematic, with symptoms incorrectly auto-labelled as "absent" even if they were repeatedly discussed. Models failed to extract many indirect descriptions of symptoms, like a patient saying they could only sleep for four hours (instead of using the word "insomnia"). [5]

LLMs are not trained to produce facts, but things which look like facts. The use of templates and rules can make them more reliable at extracting semantic information, [21] but "confabulations" or "hallucinations" (convincing but wrong output) are an intrinsic part of the technology.

Pricing

With the exception of fully open-source programs, which are free, medical scribe computer programs are rented rather than sold ("software as a service"). Monthly fees vary from mid-two figures to four figures, in US dollars. Some companies run on a freemium model, where a certain number of transcriptions per month are free. [22] [23]

Scribes that integrate into Electronic Health Records, removing the need for copy-pasting, typically cost more. [24] [ better source needed ]

Fully open-source scribes provide the software for free. The user can install it on hardware of their choice, or pay to have it installed. Some open-source scribes can be installed on the local device (that is, the one recording the audio) or on a local server (for instance, one serving a single clinic). They can typically be set not to send any information externally, and can indeed be used with no internet connection.

Notable automated medical Scribe Platforms

DeepCura AI

DeepCura AI is an AI scribe that enhances clinical workflows by automating tasks throughout the pre-visit, visit, and post-visit stages. The platform integrates the Ethereum blockchain into its scribing platform to ensure patient data integrity. [25] [26] By taking advantage of decentralized ledger, DeepCura AI aims to maintain transparent records of document creation and modification, which can be critical for compliance and auditing purposes. The company is also recognized for pioneering the concept of “ai gridhooks,” an evolution of traditional webhooks designed to facilitate complex, scalable interactions between healthcare systems. [27] [28]

See also

Related Research Articles

Information privacy is the relationship between the collection and dissemination of data, technology, the public expectation of privacy, contextual information norms, and the legal and political issues surrounding them. It is also known as data privacy or data protection.

<span class="mw-page-title-main">End-to-end encryption</span> Encryption model where only the sender and recipient can read the ciphertext

End-to-end encryption (E2EE) is a method of implementing a secure communication system where only communicating users can participate. No one else, including the system provider, telecom providers, Internet providers or malicious actors, can access the cryptographic keys needed to read or send messages.

<span class="mw-page-title-main">Electronic health record</span> Digital collection of patient and population electronically stored health information

An electronic health record (EHR) also known as an electronic medical record (EMR) or personal health record (PHR) is the systematized collection of patient and population electronically stored health information in a digital format. These records can be shared across different health care settings. Records are shared through network-connected, enterprise-wide information systems or other information networks and exchanges. EHRs may include a range of data, including demographics, medical history, medication and allergies, immunization status, laboratory test results, radiology images, vital signs, personal statistics like age and weight, and billing information.

A personal health record (PHR) is a health record where health data and other information related to the care of a patient is maintained by the patient. This stands in contrast to the more widely used electronic medical record, which is operated by institutions and contains data entered by clinicians to support insurance claims. The intention of a PHR is to provide a complete and accurate summary of an individual's medical history which is accessible online. The health data on a PHR might include patient-reported outcome data, lab results, and data from devices such as wireless electronic weighing scales or from a smartphone.

Health technology is defined by the World Health Organization as the "application of organized knowledge and skills in the form of devices, medicines, vaccines, procedures, and systems developed to solve a health problem and improve quality of lives". This includes pharmaceuticals, devices, procedures, and organizational systems used in the healthcare industry, as well as computer-supported information systems. In the United States, these technologies involve standardized physical objects, as well as traditional and designed social means and methods to treat or care for patients.

Medical transcription, also known as MT, is an allied health profession dealing with the process of transcribing voice-recorded medical reports that are dictated by physicians, nurses and other healthcare practitioners. Medical reports can be voice files, notes taken during a lecture, or other spoken material. These are dictated over the phone or uploaded digitally via the Internet or through smart phone apps.

Google Health encompasses the health and wellbeing initiatives of Google, including Fitbit and a range of other features and integrations. Google Health started in 2008 as an attempt to create a repository of personal health information in order to connect doctors, hospitals and pharmacies directly. The Google Health project was discontinued in 2012, but the Google Health portfolio re-established in 2018 before being redescribed in 2022 as an "effort" rather than a distinct division.

Protected health information (PHI) under U.S. law is any information about health status, provision of health care, or payment for health care that is created or collected by a Covered Entity, and can be linked to a specific individual. This is interpreted rather broadly and includes any part of a patient's medical record or payment history.

<span class="mw-page-title-main">Virtual assistant</span> Software agent

A virtual assistant (VA) is a software agent that can perform a range of tasks or services for a user based on user input such as commands or questions, including verbal ones. Such technologies often incorporate chatbot capabilities to streamline task execution. The interaction may be via text, graphical interface, or voice - as some virtual assistants are able to interpret human speech and respond via synthesized voices.

Clinical point of care (POC) is the point in time when clinicians deliver healthcare products and services to patients at the time of care.

A medical scribe is an allied health paraprofessional who specializes in charting physician-patient encounters in real time, such as during medical examinations. They also locate information and patients for physicians and complete forms needed for patient care. Depending on which area of practice the scribe works in, the position may also be called clinical scribe, ER scribe or ED scribe, or just scribe. A scribe is trained in health information management and the use of health information technology to support it. A scribe can work on-site or remotely from a HIPAA-secure facility. Medical scribes who work at an off-site location are known as virtual medical scribes.

Digital health is a discipline that includes digital care programs, technologies with health, healthcare, living, and society to enhance the efficiency of healthcare delivery and to make medicine more personalized and precise. It uses information and communication technologies to facilitate understanding of health problems and challenges faced by people receiving medical treatment and social prescribing in more personalised and precise ways. The definitions of digital health and its remits overlap in many ways with those of health and medical informatics.

<span class="mw-page-title-main">Braina</span> Intelligent personal assistant & dictation software

Braina is a virtual assistant and speech-to-text dictation application for Microsoft Windows developed by Brainasoft. Braina uses natural language interface, speech synthesis, and speech recognition technology to interact with its users and allows them to use natural language sentences to perform various tasks on a computer. The name Braina is a short form of "Brain Artificial".

Babylon Health was a digital-first health service provider that combined an artificial intelligence-powered platform with virtual clinical operations for patients. Patients are connected with health care professionals through their web and mobile application.

<span class="mw-page-title-main">Artificial intelligence in healthcare</span> Overview of the use of artificial intelligence in healthcare

Artificial intelligence in healthcare is the application of artificial intelligence (AI) to analyze and understand complex medical and healthcare data. In some cases, it can exceed or augment human capabilities by providing better or faster ways to diagnose, treat, or prevent disease.

<span class="mw-page-title-main">Otter.ai</span> Transcription software company

Otter.ai, Inc. is an American transcription software company based in Mountain View, California. The company develops speech to text transcription applications using artificial intelligence and machine learning. Its software, called Otter, shows captions for live speakers, and generates written transcriptions of speech.

<span class="mw-page-title-main">Merative</span> U.S. healthcare company

Merative L.P., formerly IBM Watson Health, is an American medical technology company that provides products and services that help clients facilitate medical research, clinical research, real world evidence, and healthcare services, through the use of artificial intelligence, data analytics, cloud computing, and other advanced information technology. Merative is owned by Francisco Partners, an American private equity firm headquartered in San Francisco, California. In 2022, IBM divested and spun-off their Watson Health division into Merative. As of 2023, it remains a standalone company headquartered in Ann Arbor with innovation centers in Ireland, Hyderabad, Bengaluru, and Chennai.

Proton AG is a Swiss technology company offering privacy-focused online services. It was founded in 2014 by a group of scientists who met at CERN and created Proton Mail.

Toloka, based in Amsterdam, is a crowdsourcing and generative AI services provider.

Artificial intelligence in mental health is the application of artificial intelligence (AI), computational technologies and algorithms to supplement the understanding, diagnosis, and treatment of mental health disorders. AI is becoming a ubiquitous force in everyday life which can be seen through frequent operation of models like ChatGPT. Utilizing AI in the realm of mental health signifies a form of digital healthcare, in which, the goal is to increase accessibility in a world where mental health is becoming a growing concern. Prospective ideas involving AI in mental health include identification and diagnosis of mental disorders, explication of electronic health records, creation of personalized treatment plans, and predictive analytics for suicide prevention. Learning how to apply AI in healthcare proves to be a difficult task with many challenges, thus it remains rarely used as efforts to bridge gaps are deliberated.

References

  1. 1 2 3 4 5 6 7 8 9 10 11 12 "RACGP - Artificial intelligence (AI) scribes". www.racgp.org.au. Retrieved 13 December 2024.
  2. 1 2 3 4 5 "Researchers say an AI-powered transcription tool used in hospitals invents things no one ever said". AP News. 26 October 2024. Retrieved 13 December 2024.
  3. 1 2 3 4 5 Kuzub, Alena (1212). "How will AI scribes affect the quality of health care?". Northeastern Global News. Retrieved 13 December 2024.
  4. 1 2 Adrian, David; et al. (October 2015). "Imperfect Forward Secrecy: How Diffie–Hellman Fails in Practice" (PDF). Archived (PDF) from the original on 2015-09-06.
  5. 1 2 3 4 van Buchem, Marieke M.; Boosman, Hileen; Bauer, Martijn P.; Kant, Ilse M. J.; Cammel, Simone A.; Steyerberg, Ewout W. (26 March 2021). "The digital scribe in clinical practice: a scoping review and research agenda". npj Digital Medicine. 4 (1): 57. doi:10.1038/s41746-021-00432-5. ISSN   2398-6352. PMC   7997964 . PMID   33772070.
  6. ↑ "NHS plans leave 'anonymous' medical data vulnerable". New Scientist. Retrieved 13 December 2024.
  7. 1 2 3 Dorn, Spencer. "Where AI Ambient Scribes Are Heading". Forbes. Retrieved 13 December 2024.
  8. ↑ Stanton, Rich (22 October 2021). "Google AI department sued for using the health data of 1.6 million NHS patients | PC Gamer". PC Gamer. Archived from the original on 22 October 2021. Retrieved 13 December 2024.
  9. ↑ "ClinicianFOCUS/FreeScribe". ClinicianFOCUS. 12 December 2024. Retrieved 13 December 2024.
  10. ↑ Schmidt, Rebecca (12 February 2024). "Automatische Transkriptionssoftware – ein Erfahrungsbericht". Sozialwissenschaftliche Methodenberatung (in German). doi:10.58079/vsz4 . Retrieved 13 December 2024.
  11. ↑ Karopka, T.; Schmuhl, H.; Demski, H. (2014). "Free/Libre Open Source Software in Health Care: A Review". Healthcare Informatics Research. 20 (1): 11–22. doi:10.4258/hir.2014.20.1.11. PMC   3950260 . PMID   24627814.
  12. ↑ Perera, Chandrashan (1 June 2012). "Principles of Security for the use of Mobile Technology in Medicine". Journal of Mobile Technology in Medicine: 5–7. doi:10.7309/jmtm.10.
  13. ↑ "How Can You Protect and Secure Health Information When Using a Mobile Device? | HealthIT.gov". www.healthit.gov. Retrieved 13 December 2024.
  14. ↑ Martínez-Pérez, Borja; de la Torre-Díez, Isabel; López-Coronado, Miguel (January 2015). "Privacy and Security in Mobile Health Apps: A Review and Recommendations". Journal of Medical Systems. 39 (1): 181. doi:10.1007/s10916-014-0181-3. PMID   25486895.
  15. ↑ Xing, Yawen; Lu, Huizhe; Zhao, Lifei; Cao, Shihua (9 September 2024). "Privacy and Security Issues in Mobile Medical Information Systems MMIS". Mobile Networks and Applications. 29 (3): 762–773. doi:10.1007/s11036-024-02299-8.
  16. ↑ The Nizza Secure-System Architecture (PDF) (Report). 2005. Retrieved 13 December 2024.
  17. ↑ WEiss, Luca. "Porting mainline Linux to mobile phones". archive.fosdem.org. Retrieved 13 December 2024.
  18. 1 2 3 4 Edwards, Benj (28 October 2024). "Hospitals adopt error-prone AI transcription tools despite warnings". Ars Technica. Retrieved 13 December 2024.
  19. ↑ Landi, Heather (11 October 2024). "Abridge integrates UpToDate decision support into AI scribe". www.fiercehealthcare.com. Retrieved 13 December 2024.
  20. ↑ "RACGP - Three key principles for the secondary use of general practice data by third parties". www.racgp.org.au. Retrieved 13 December 2024.
  21. 1 2 Edwards, Benj (18 April 2023). "GPT-4 will hunt for trends in medical records thanks to Microsoft and Epic". Ars Technica. Retrieved 13 December 2024.
  22. ↑ Brodwin, Erin (21 March 2024). "Comparing AI medical scribes by price and features". Axios. Retrieved 13 December 2024.
  23. ↑ "AI scribe wars heating up - Medical Republic". 2 November 2024. Archived from the original on 2 November 2024. Retrieved 13 December 2024.
  24. ↑ "AI medical scribes: Boosting efficiency or risking over-reliance?". KevinMD.com. 29 October 2024. Retrieved 13 December 2024.
  25. ↑ "Charting the Future: The Role of AI in Transforming Nursing Documentation". pmc.ncbi.nlm.nih.gov. Retrieved 27 January 2025.
  26. ↑ "Artificial intelligence in your office". www.mdedge.com. Retrieved 27 January 2025.
  27. ↑ "Embracing Artificial Intelligence: Revolutionizing Nursing Documentation for a Better Future". pmc.ncbi.nlm.nih.gov/. Retrieved 27 January 2025.
  28. ↑ "AI Clinicat Automation Platform". deepcura.com. Retrieved 27 January 2025.
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.