Assuta, Microsoft develop AI medical assistant to reduce relieve doctors' documentation tasks

Assuta Medical Centers and Microsoft are developing an artificial intelligence system that will effectively serve as a personal assistant for doctors. The AI will collect relevant medical information, summarize test results and previous documents, document meetings with the patient, and automatically draft visit summaries. 

The AI is not intended to replace the doctor, but to relieve them of documentation and search tasks, leaving them with more time for examinations, conversations, and medical decision-making.

Unlike standard medical computer systems, in which doctors must move between screens, tests, hospitalization summaries, and referral letters themselves, the AI agent is intended to read the medical record and present important information ahead of the meeting. 

Before the patient enters the room, the doctor will be able to receive a summary of previous diagnoses, medications, recent tests, treatments that have already been tried, and changes in the patient’s medical condition. The goal is to reduce the time spent locating information within large volumes of documents and lower the risk that a significant detail will be overlooked in the file.

During the visit, the system will listen to the conversation between the doctor and the patient, after the necessary approvals are obtained and in accordance with privacy rules, to produce an organized transcript. It will then process the information and draft a complete medical document that includes the patient’s complaints, medical history, examination findings, the doctor’s assessment, and recommendations for continued care. 

The doctor will be required to review, correct, and approve the document before it is added to the medical record. Assuta stressed that the system will serve only as a decision support tool and will not replace medical judgment. Responsibility for the diagnosis, the selection of tests, and treatment will remain with the doctor.

The system is also intended to assist after the visit. According to the plan, it will be able to identify when a patient requires a test or an additional appointment, locate a suitable time, and help schedule a follow-up visit based on medical need and availability in the system. Additional functions may be introduced in the future throughout the continuum of care, subject to medical oversight and information security regulations.

AI system designed to shorten visits without reducing time of doctor-patient conversations

One of the main challenges facing the healthcare system is the amount of time doctors spend on documentation, typing summaries, reviewing documents, and carrying out administrative tasks. In some cases, this work continues even after the clinic day has ended. According to the targets set for the project, use of the agent could save hundreds of doctor work hours each month.

These time savings could make it possible to shorten visits without reducing the amount of time doctors spend speaking with patients.

Instead of looking at a screen and typing details of the conversation during much of the appointment, the doctor would be able to rely on the system to prepare the documentation. This could allow more patients to be seen during the workday, increase appointment availability, and reduce waiting times at clinics and medical institutes.

Developing such a system requires more than simply connecting transcription software to a doctor’s computer. To understand medical documents, the agent must undergo an extensive process of learning, practice, and training. As part of that process, the systems are provided with models of medical documents, visit summaries, test results, and professional terminology, while identifying details are removed and patient privacy is protected.

The system is trained to distinguish essential information from minor details, identify connections between diagnoses, medications, and tests, and draft concise summaries without omitting important information. Medical teams review the results, flag errors, and teach the system how to present information in a way that will assist doctors in their work.

One of the main challenges will be preventing errors commonly associated with artificial intelligence systems, including the omission of details, confusion between patients, or the creation of information that did not appear in the original documents. For that reason, the system’s output will not be entered automatically into the medical record without human review. Development and implementation will be carried out gradually, while examining the system’s accuracy, safety, and effect on the work of medical teams.

The first pilot is expected to begin within a year. The system will likely initially operate on a limited scale so that Assuta can examine how it integrates into daily work, how much time it saves, the quality of the summaries, and how comfortable doctors and patients feel with the automatic documentation of appointments.

Microsoft to develop more AI agents 

Alongside the development of the personal agent, Assuta and Microsoft agreed to gradually transition to the company’s cloud infrastructure and establish an organizational platform for developing and operating additional artificial intelligence agents. The infrastructure could eventually be used for solutions in medicine, service, and operations, while maintaining strict rules on privacy, information security, and the protection of medical information.

Sarit Sass, healthcare organizations director at Microsoft Israel, said: “We believe that the future of the healthcare system relies on combining the human expertise of medical teams with the capabilities of artificial intelligence.” She said the goal of the partnership is to allow medical professionals to focus on caring for the patient.

Ofir Shani Shaharbani, head of Assuta’s Information Systems and Technology Division, said the partnership was “a significant milestone in our journey.” He said the new infrastructure would make it possible to bring artificial intelligence to the heart of medical activity “in a responsible and secure manner, while strictly protecting patient privacy.”