When AI and Medicine Meet

By: Mira Postelnek  |  September 20, 2021

By Mira Postelnek 

With each passing day, technology gets progressively more advanced. Our everyday lives have become impacted by the incredible technological advances being made in the world. A prime example of this can be seen in the world of applying artificial intelligence (AI) to medicine. 

It has become increasingly apparent that digital scribes can be utilized to create a more efficient and productive work environment across the entire medical field. Using AI to document medical information instead of having a person manually inputting the information could cut back on clinician workload significantly. Notes could be written by AI scribes while clinicians focus on their patients. The use of digital scribes would also be eliminating the need for doctors to split their attention between the patient and computer, which would increase patient satisfaction, enhance the patient-doctor connection, and decrease clinical burnout (since the majority of clinical burnouts are associated with tedious clinical documentation). 

Additional benefits of AI technology were found in a recent study done by McKinsey & Company. The study hypothesized that, if done right, AI integration into the medical world will lead to “increased flexibility and scalability, with the ability to operate 24/7 and scale up or down with demand. Improved quality, from spot-checking to 100 percent quality control.” Not only would digital scribes bring quality and flexibility, but introduction of digital scribes could also increase the rate of productivity and reduce tasks from days to minutes. 

There are also clear economic advantages that come with implementing AI technology into the medical world. Data collected by AI technology could be used to improve behavioral understanding, potentially predict healthcare trends, and save up to 20% of current work efforts. AI has the potential to create improvements in all areas of healthcare, from diagnosis to treatment. 

While AI has its productive uses in the medical world, there are a number of issues that could arise when AI technology is used for digital scribing. In a recent study by The National Health and Medical Research Council, it was found that AI transcriptions of human conversations are not always accurate: “word-error-rate of automated speech recognition (ASR) engines was 35% or higher. These are best-case scenario results for current ASR technologies.” There needs to be better statistics if AI will help, rather than hinder, efficiency and productivity in the medical world. Additionally, these statistics don’t account for the various environmental factors that can interfere with the accuracy of a digital scribe, such as difficulty picking up a conversation if not angled at exactly the right position, or AI’s difficulty with recognizing and understanding causal conversational cues. Since doctor appointments often consist of spontaneous side conversations and unrelated topics, the computer would get flustered and the resulting transcript would be flawed and confusing. Faulty documentation of important medical data can create more work for the clinician who has to edit the transcription later on. 

Additionally, the notes taken by a medical scribe should reflect the specific questions, observations, physical examination, and sometimes non-verbal conclusions performed by the clinician. These details are practically impossible to capture during an AI examination unless a human clinician specifically verbalizes them during a consultation. Furthermore, it is not uncommon for a clinician to change their opinion post-evaluation or revise previous observations. This change would be difficult to integrate into the automated summarization code produced by a digital scribe and would require more sophisticated technology to accomplish. 

Digital scribes lack the critical thinking and structured thought process, involved in a clinician’s manual notes. Many of the advocates for the integration of digital scribes overlook  the complexity of the sociotechnical healthcare system. When evaluating AI integration into the medical world, it is crucial to take into account, “effective quality care, patient satisfaction, clinician efficiency, documentation time, and organizational dynamics within a clinic.”  There needs to be thorough research conducted between patients, clinicians, and AI technology before installing this design  into systems. These advancements have the potential to change clinician-patient communication; supporting clinicians to engage with their patients. However, without more research, the digital scribe solution could lead to detrimental results, creating system delays, and misdiagnosis. 

There are various challenges to the integration  of digital scribes into the medical world. However, those researching the matter should explore solutions to these challenges to develop  an advanced system to overcome these deficiencies. One suggested solution is to gather additional data to improve AI software to make it more accurate and efficient. Another possible solution would be to form a symbiotic human-AI relationship that could improve the quality of patient care, instead of aiming to have digital scribes replace human ones all together. With tremendous daily technological advancements and efforts to collect medical data, the AI medical scribes could eventually transform the way patients and clinicians interact to create a more connected, engaged environment. 

*** ***