top of page

Tomorrow’s doctors don't see AI as an option, but as part of good practice

Writer: Monsana Team
Monsana Team
Nov 10
4 min read

It is the year 2032. A doctor overlooks an AI warning based on a lab report, or a radiologist distrusts an AI tool that flags a tumour on a scan. We put these and other scenarios to 45 bachelor’s and master’s students. Would they consider it bad clinical practice not to use these new technologies?


Monsana conducted the survey during a presentation at the Junior Orsi Academy, asking a mix of medical and engineering students for their views on the adoption of AI.


Technology adoption in healthcare is often slow, sometimes for good reason. The stakes are high, and it is not always clear whether a new technology will improve the current standard of care. Doctors’ capabilities, patients’ conditions, tolerance for risk, trust, readiness to use technology and the economics behind it all vary. Each factor can influence whether a new tool adds enough value to the process.


We wanted to understand what the next generation of professionals thinks about the use of AI and other innovative solutions in daily practice.


The view of future professionals about AI in healthcare

The results suggest that students largely accept the use of AI and other technology to support diagnosis and treatment decisions. In those scenarios, more than 80% of the students felt that clinicians could be blamed for not using the technology.

Individual cases are detailed in the appendix. Key question: Was this a fair clinical decision or a medical error for which the doctor is responsible?
Individual cases are detailed in the appendix. Key question: Was this a fair clinical decision or a medical error for which the doctor is responsible?

When it comes to performing surgery, respondents felt that more should be left to doctors’ clinical judgment. Even so, the majority still believed that technology should play a role in assisting or supervising the surgical team.


One scenario that is already common today received a different response: only 11% of students agreed that an analysis made by ChatGPT and brought to a physician by a patient should be taken seriously. This suggests that students see doctors as being responsible for deciding when and how to use AI tools, rather than feeling obliged to act on a patient’s request.


This clear contrast shows established tools within a hospital’s workflow may be much more acceptable than a suggestion brought in by a patient.


When will AI become part of good practice?

Respondents had different views on timing. Fourteen said AI is already part of good clinical practice, 10 believed it would become so within five years, and 19 chose within ten years. Only one believed that AI would not be needed to be a great doctor.

"When will AI become part of good clinical practice?"
"When will AI become part of good clinical practice?"

As AI tools become more common, healthcare professionals will need to decide when they add enough value to be used, and how to integrate their output into clinical workflows alongside their own judgment.


Looking further ahead, a controversial question could be raised: when might a patient sue a doctor for not using an available AI tool? Even if such a legal case were unsuccessful, it could prompt debate about what patients can reasonably expect. Standards of care tend to change slowly, with a roughly 17-year lag often cited between a clinical development’s emergence and its becoming a mainstream expectation.


What do you think lies ahead? Would the results be different if we conducted the same survey among doctors aged 50 and over? 


Appendix: Cases and survey results

Respondents were asked whether each scenario was a case of “blame” or a “fair call.” 


Case 1. The blood pressure monitor

A patient visits their GP with severe headaches and blurred vision. The patient attributes these recurring symptoms to stress. The doctor suspects migraine and decides not to measure the patient’s blood pressure, despite having a monitor available.

The patient is later diagnosed with a hypertensive crisis.

Case 2. The surgical instrument counter

A camera system in the operating room tracks the instruments and surgical sponges used during the procedure and alerts the team if anything is missing. The surgeon switches it off because he is uncomfortable with his operations being filmed.

A few days later, an instrument is discovered inside the patient’s body.

Case 3. The tumour missed on the scan

A radiologist examines a scan and finds no signs of a tumour. The hospital has AI software that can act as a second reader and flag suspicious abnormalities. The radiologist decides not to use it.

Months later, a tumour is discovered, and retrospective analysis shows that the AI would have flagged it.

Case 4. The scanner in the operating room

A surgeon removes a breast tumour. The hospital has an intraoperative imaging system, such as the XEOS AURA 10, which can examine removed tissue during surgery. The surgeon considers conventional assessment of surgical margins sufficiently reliable and wishes to avoid additional operating time.

Months later, residual tumour tissue is discovered.

Case 5. The surgical robot

A patient undergoes a complex operation. The surgeon has extensive experience with conventional techniques, while the hospital also has an advanced surgical robot. Although the robot offers greater precision, the surgeon finds the conventional approach faster. During surgery, an incorrect incision causes permanent damage. It is established that robotic surgery would have avoided the mistake.

Case 6. The warning on the lab result

The hospital uses an AI system to flag possible diseases and recommend further testing based on laboratory results. In this case, the system flagged possible impaired kidney function due to elevated creatinine and reduced eGFR. However, the doctor is unwilling to use the AI system and therefore does not review its report.

Weeks later, chronic kidney disease is diagnosed, and earlier intervention could have slowed its progression.


Case 7. Doctor ChatGPT

A patient presents their GP with persistent fatigue, joint pain and unexplained weight loss, and ChatGPT has already suggested to the patient a possible autoimmune disease. The doctor does not want to hear the AI suggestion and attributes the symptoms to stress.

Months later, further tests confirm the autoimmune disease, which has since progressed.


Case 8. The patient with no options left

A patient has been through all standard treatments. The doctor says nothing more is possible.

The hospital has an AI system that can find relevant clinical trials and innovative treatments. The doctor does not use it and does not discuss this with the patient.



Comments


Monsana-logo
bottom of page