An AI physician? Not quite yet

Dr. Adam Cifu raises doubts on Artificial Intelligence's ability to understand patient histories and highlights the challenges of doctor/patient communication.

What can Artificial Intelligence do in medicine?

The field of Artificial Intelligence (AI) in medicine is experiencing rapid developments, with numerous applications that could radically transform medical practice in the coming years.

AI is showing remarkable progress in medical diagnostics, especially in the analysis of images such as X-rays, tomographs and MRIs. Advanced algorithms can detect early signs of disease, improving the timeliness of diagnosis and enabling early intervention. The use of genetic data and biomarkers further enriches the diagnostic capabilities of AI.

In the therapeutic field, AI is already being used to propose evidence-based treatment protocols. The analysis of clinical data enables the identification of optimal therapeutic strategies for specific conditions. A crucial aspect of AI in medicine is its ability to predict disease risk before symptoms appear. By analysing anamnestic, genetic and environmental data, AI can identify individuals at risk and enable targeted preventive interventions. This could lead to a paradigm shift towards predictive and preventive medicine.

Parallel to technological developments, important ethical issues arise. Patient privacy, data security, and the interpretation of AI decisions are critical issues that require a careful approach and appropriate regulations. Furthermore, the increasing presence of AI in medical practice raises questions about its social acceptance by both patients and physicians.

AI still does not understand fundamental details in medicine

In a recent article, Dr Adam Cifu raised important thoughts on the current state of Artificial Intelligence in medicine, highlighting both the potential and challenges associated with this technological advancement. Dr Cifu recognises AI as a potential ally in improving diagnostic performance. AI can analyse large amounts of clinical data, medical images and genetic information in much shorter timescales than human professionals. This can lead to more timely and accurate diagnoses, especially in cases where speed is of the essence.

However, the author points out that current AI may have limitations in interpreting the complexity of individual medical histories. The ability to discern nuances such as pain tolerance, psychological context, and the patient's approach to symptomatology may still be beyond the scope of current AI.

In the area of therapeutic decision-making, AI shows great potential. Its ability to analyse huge datasets can provide precise recommendations for standardised treatments. However, Dr Cifu emphasises that when it comes to more complex decisions, such as the choice of a personalised treatment plan, AI may find it difficult to integrate unique individual factors that today only a human physician could recognise and assess.

The AI-patient relationship in 2023

One of Cifu's main concerns is the doctor-patient interaction. He expresses doubts about the ability of patients to accept medical advice from an artificial intelligence system. The lack of empathy and understanding of emotional details could make this form of communication less effective or even unacceptable for many patients.

The author shares concrete examples of interactions between patients and AI, demonstrating the current limitations in AI-generated responses. These examples highlight the need for further developments to improve the adaptability of AI to different facets of medical demands:

"Doctor, my partner, John, was admitted to hospital. They did his initial tests. A heart attack, which fortunately it seems not so extensive. I would need to discuss the situation with you. The doctors have spoken to me, but I would need to better understand what is going to happen next".

Faced with this message, Dr. Cifu reports this reply from the AI:

"I am sorry, but I am unable to answer your request because it does not contain a specific question or concern. If you have health questions or would like to discuss something, please do not hesitate to make an appointment. You can do so by following this link: [click here to view or make an appointment]".

Dr. Cifu's response to the message from his patient's partner was a phone call.

Artificial Intelligence: much room for improvement

Dr Cifu is optimistic about the future, acknowledging that AI is still in a developmental stage. In conclusion, while AI offers considerable potential to improve medical practice, it is important to address the critical challenges raised by practitioners like Dr Cifu. A balanced approach that recognises the benefits of AI but also its current limitations is essential to guide future development in a responsible manner that is geared towards effectively improving patient care.

  1. Cifu A. The AI Doctor will message you now (if you'd like). Sensible Medicine. Dec 20 2023.