Using Artificial Intelligence (AI) tools for record management in doctor consultations

Medical transcription software uses speech to text transcription methods to convert audio files of doctor and patient conversations into a written record in real time. Using various technologies and algorithms such as terminology dictionaries/libraries, coding assistance and search capabilities, a written record is created – this record may not be a word for word transcript but an organised summary of the clinical features of the discussion. The record can be transferred into the electronic medical record (EMR) for the doctor to use.

Use of AI applications in note taking may provide efficiencies for doctors and practices.

• Accuracy: conversations are recorded or streamed and a written record is created following the consultation, providing potential for accurate data collection.
• Efficiency: Reduction in the need to manually type or outsource transcribing services may result in increased time efficiency, potentially allowing more time for other activities.
• Cost savings: may reduce time and costs associated with transcribing notes once the cost of the AI tool is taken into consideration.
• Security: encryption and redaction capabilities to protect confidential patient information may be available.
• Reduced losses due to litigation: notes are completed for each consultation for review by the doctor, reducing legal risks associated with not having accurate records on file, or records not kept at all.
• Links with EMR: the written record can be linked directly to the patient’s EMR.
• Patient satisfaction: the consultation is available in real time (at the conclusion of the care episode) if the patient should wish to be provided with a copy. The doctor can spend the consultation focused on the patient, rather than the computer screen and keyboard.

For doctors adopting or considering implementing AI assisted technology, the following potential limitations should be considered and managed appropriately.

• New system set up: Ensuring that it is compatible with practice systems; and developing processes/ protocols and staff training.
• Accuracy/interpretation of data: The record will require review to ensure it remains an accurate summary of the consultation. Inaccuracy may arise from:
  • Poor transcription e.g. accent, dialect, idioms, slang – of doctor or patient(s)
  • Misattribution of comments to individual parties such as confusing comments from patient vs doctor vs attendees at consultation
  • Misunderstanding of subject and context, the written record may miss or focus on the wrong information relative to the patient’s situation or need
  • Data/content loss as the software reduces a written record to a summary, potentially missing clinically relevant material
  • Misclassification of information by incorrectly grouping history; symptoms; findings; plan
  • Managing complex and drawn-out consultation data.
• Transfer of record errors: There may be a risk of data loss or incorrect data storage when transferring the record from the AI system to your medical records system. The AI record notes may not have patient identity data recorded with the note, meaning that delayed transfer can make it more difficult to identify which patient they belong to (other than by time of generation).
• Format of notes: Different AI tools may organise notes differently – eg the SOAP format suits short focused consultations, but might not be appropriate for a consultation where a full medical history is being taken (long case format). There are likely to be specific advantages and disadvantages to a particular format.
• Consent from patients: Not all patients will agree to use an AI record system or trust in its capabilities, hence a mixed model approach to note taking may be required. Recording of private discussions without consent is a criminal offence in some jurisdictions e.g. NSW; South Australia; Western Australia; and the ACT.
• Limited use in Australia: Hence limited experiences to draw on in relation to its implementation and approval.
• Cost: Be sure to understand all costs involved in set up, roll out and ongoing fees.

Some of the medico legal risks to be aware of include:

• Accuracy of the notes. The doctor conducting the consultation is responsible for the accuracy of the medical record of that consultation. Any patient notes generated are deemed to be signed off/approved by you. Before entering the AI-generated record into the clinical record, the doctor must check and if necessary, edit the document to ensure accuracy, and that relevant content has been included (or not been excluded).
• Consent: Gaining consent from the patient before recording is critical and should be documented in the patient’s medical record. Providing information and signs in the practice about the AI technology, while helpful for expectation setting, does not substitute for consent. Have a system in place to seek and record the consent for the use of this technology (it may be inbuilt).
• Security of content: Determine whether the content is encrypted/redacted and if it is stored (even temporarily) on an overseas server (as this may breach Australian privacy legislation if specific consent is not obtained).
• Privacy/Storage: you will need to ensure any provider meets the obligations under the Privacy Act before utilising their services. The contract between the AI program provider and the user/doctor should cover information security and Australian privacy law. Is the AI program regulated by the TGA. and if so, approved by them as part of their regulation of software based medical devices?

There are many components to consider when using AI in record management. Being comfortable with the technology and recognising that it is a tool to assist, and not replace, your practice is important. There are a range of other factors that come into consideration such as cost, security, privacy, and reliability. MDA National has prepared a checklist to assist in this process when comparing potential vendors to help achieve your goals.

At present, there are limited guidelines around the use of AI in medical documentation. Take time to read some of the links below to further develop your understanding of medico legal risks that can arise through the use of AI based technology.

• As a practice owner: There are no obligations to use the technology. Use the checklist provided to help determine if AI note taking technology fits in with your business plans.
• As an employee: If your employer, or the practice you work at implements AI technology in note taking, you may be required to use it. Be sure that adequate training is provided and clear policies and procedures are in place, along with any legal requirements of the employer/practice owner and yourself.
• As a contractor: You likely have a right to choose whether you use an AI record system or not as an independent contractor (subject to your contractual agreement with the practice). Be sure that adequate training is available and clear policies and procedures are in place, along with any legal or contractual obligations that either the contractor or practice must abide by.

Some AI based medical record systems provide ‘record only’ functionality. This is simpler to manage from a compliance perspective, as the relevant regulatory framework will focus more on confidentiality / privacy / consent and record content requirements.

There may also be additional AI assisted functionality available (such as AI assisted differential diagnosis; AI assisted prescribing; AI assisted billing; AI assisted form completion). Additional functionality may raise complex ethical, professional, regulatory and indemnity issues for which relevant policies and guidelines are lacking at present time. Users need to be aware that they may be adopting functionality for which the implications are not well understood.