Many countries around the world face a significant healthcare challenge. Whenever a citizen wants to see a specialist, whether it’s an ophthalmologist, a physiotherapist, a dermatologist, or undergo a medical imaging procedure, the waiting times can stretch for months or even a year. As a result, by the time the individual gets the examination, they may either have already recovered or reached a terminal stage of their condition. It’s a despairing situation, and there seems to be no immediate improvement in sight.
Perhaps in the long run, artificial intelligence (AI) could help alleviate this problem, especially through a project called Doctor Dignity. This project involves a large language model (LLM), which is open source and capable of passing the U.S. medical license exam.
Isn’t that incredible?
This software utilizes Meta’s Llama2, fine-tuned with medical terminology, and can be used on iOS, Android, or as a web version. The advantage is that this pocket doctor is free, instantly accessible, and capable of maintaining medical confidentiality. However, it’s unclear where the project stands at the time of this article’s publication, and we’re still far from having something easily usable by the general public.
Of course, it’s still a long way from something one would fully trust with their health or life (although, in some cases, the care provided by certain medical interns in emergency situations isn’t much different), but considering that Doctor Dignity can pass the U.S. medical license exam, one can assume it has some knowledge in the field. Importantly, it can also learn from human feedback, refining its responses over time.
This marks just the beginning of the “virtual doctor,” perhaps a tool capable of diagnosing 90% of common minor ailments, thereby relieving real healthcare professionals. Or, who knows, even replacing them if they prove to be less effective than AI. In any case, having such an application accessible at any time for any medical question could bring relief to even the most hypochondriac among us.