Benefits and Ethical Considerations of Using AI for Predicting Incapacitated Patients' Consent

Share this content:
The use of an artificial intelligence algorithm could become valuable for medical decision-making processes.
The use of an artificial intelligence algorithm could become valuable for medical decision-making processes.

An artificial intelligence (AI)-driven autonomy algorithm comprising data mining electronic health records (EHRs) and social media is becoming an area of increasing interest for predicting consent in patients unable to make healthcare decisions.

According to the authors of a viewpoint article published in the AMA Journal of Ethics, the use of this algorithm could become valuable for "medical decision-making processes, augmenting the capacity of all people to make health care decisions in difficult situations."

The use of AI for data mining EHRs and social media is not a new concept; however, recent interest has increased over the utility of this practice for predicting consent for medical decisions in incapacitated patients. Data gained from EHR and social media profiles include past patient requests, comments, and discussions regarding potential health wishes. When making life-and-death decisions for incapacitated patients, both clinicians and surrogates, use of an AI algorithm can be an effective strategy for managing care and making decisions more in line with patient preferences.

An AI-driven autonomy algorithm may also analyze the EHRs of incapacitated patients with a specific disease and, while adjusting for key demographic factors, can predict an individual patient preference to a treatment option. Combining these data with the individual's social media behaviors, including comments, likes, group participation, and replies, may further improve prediction of the patient's medical wishes, perhaps even more so than the patient's spouse or designated surrogate.

Although this system has merit and potential, there are specific limitations that should be highlighted. First, the algorithm may reflect existing biases, with certain treatment choices being chosen on the basis of the practicing physician's specialty. In addition, patients with a lower health literacy who previously showed affinity to less effective health interventions on social media and/or in their EHRs can reduce the practicality of some predictions.

"It is concluded that an AI-assisted autonomy algorithm, if thoughtfully implemented and judiciously used, could offer some relief from the aforementioned triple burden posed by incapacitated patients: it could lead to improved respect for autonomy, reduced burnout of surrogates, and economic gains for society," the authors wrote. "However, we must tread carefully in the implementation of the proposed technology and remember that algorithms function as decision aids, not dictates."

Reference

Lamanna C, Phil MM, Byrne L. Should artificial intelligence augment medical decision making? The case for an autonomy algorithm. AMA J Ethics. 2018;20(9):E902-E910.

Free E-Newsletter