According to a report in The Wall Street Journal, DeepScribe, a company specialising in utilising artificial intelligence as a medical scribe, has employed 200 human contractors to review medical visit recordings and correct errors made by the AI.
The medical scribe platform developed by DeepScribe is currently employed by nearly 1,000 physicians and healthcare providers. It serves to record, transcribe, and generate physicians' notes based on conversations between healthcare professionals and their patients during medical visits. However, the AI algorithms used in the platform are not without issues. They sometimes produce inaccurate information, including errors in listing medications and generating "disjointed, nonsensical sentences."
To address these shortcomings, each note produced by the AI is scrutinised by one of the human contractors, whose primary task is to identify and correct errors, especially in medication information and ICD-10 code generation. The reviewers possess only minimal ICD-10 training and rely on the physician and medical staff to verify codes in the reports.
The involvement of human workers has significantly improved the accuracy of the AI, increasing it by 15 percentage points, resulting in a 95 percent accuracy rate, as reported. However, this use of human contractors also raises ethical concerns, as the recorded interactions include sensitive patient information, such as patients' names and intimate details about their lives.
Despite these challenges, DeepScribe has experienced substantial financial success. Since its establishment in 2017, the company has secured $37.3 million in venture capital funding. It is projected to achieve $6.5 million in revenue this year, and its valuation reached $180 million in 2021.
Click here to read the original news story.