What a Family Medicine Physician thinks about ML in healthcare
A summary of my interview with a Family Medicine Physician. This is one of my 18 interviews with clinicians for my MPH capstone (link here) for UC Berkeley’s School of Public Health.
Visit my Building Trust and Adoption in Machine Learning in Healthcare site (link here) for the abridge and full version of my MPH capstone as well as upcoming summaries of interviews and additional research.
Note that this interview is a de-identified summary for ease of reading and privacy. It has been approved for distribution by the interviewee.
“I have been working in the field for so long, so it is hard for me to have the time or capacity to understand how things like ML work.”
Job Background
I am a Family Medicine Physician and am part of a four-doctor private practice, focusing on Internal Medicine and Geriatrics. The majority, at least 75%, of my patients are elderly folks over 65 years old. Many have hypertension, diabetes, heart disease, COPD, and now dementia. I intend to practice for another five to ten years.
Familiarity with ML in healthcare
To start off, what have you heard about artificial intelligence and/or machine learning in healthcare?
I haven’t heard a whole lot. I heard that people are looking into ML that can learn how to read X-rays and pathology slides.
Past and future use
Have you used any ML tools? Would you?
No, I haven’t, but I would. I am typically someone who does adopt technology. For example, we transitioned from paper to the EHR pretty early.
Excitement and concerns
What about ML in healthcare is concerning or exciting for you? What else is exciting or concerning for you?
I am not really concerned. I think ML will be complementary to what we do as clinicians instead of competitive. We can see that technology has often been complementary in different fields.
Yet, we need to make sure that we don’t completely rely on ML and lose the human skill of medicine. When you use more technology and less of your brain, then you lose your skills. Take GPS as an example: We used to know all the roads and how to get places. Now, we are dependent on the machines, and no one has any sense of direction. We need to make sure our medical brains are stimulated and kept strong.
I did hear a talk once about ML over-reading radiology scans and having too many false positives. So that may be a little concerning.
Ethics and privacy
How does privacy fit into all of this?
Everything has gone digital nowadays. So, privacy will be an issue with any new technology.
How should the data be used? Who should or should not have access to it?
The access to the data should be available to those who provide direct care to the patient and any agency that would use the data to improve the quality, efficiency, and price of the ML tool.
ML knowledge and model explainability
At what level do you need to understand how the model makes its prediction?
I need to know about ML at a more basic level than other technologies. I have been working in the field for so long, so it is hard for me to have the time or capacity to understand how things like ML work. For example, I don’t think many clinicians of my age truly understand how MRI machines work. We know that there is a magnet and it takes images. We know what to look for in the images and what diagnoses to catch, but we don’t know the details of how we get the image.
External validation needs
What types of government and/or non-government institutions would play a role [in external validation]?
Yes, I think there would be a lot of oversight. That is important when you are doing diagnosis and treatment of patients. I think the government needs to be the ultimate authority. It may assign the task to private companies, but again the government needs to take on the responsibility.
Implementation
When an ML tool gets implemented, how should that be done? Who should have access first; who should not?
These ML tools still need to be individually supervised by humans. Like when someone is working under you for a while. You mentor them until they learn and you feel comfortable giving them space. The same goes for ML. Different people have different comfort levels with letting go, so you need to know that too.
Buying process
Where you practice medicine, who all would be involved in making the decision to purchase and use an ML tool?
I would have to look to see if this tool were relevant to my practice first. Then if I thought it were helpful, my clinical partners would evaluate it. We would need to see if it delivers healthcare better and if it is financially viable. It needs to make sense for a small practice like ours. So, decisions like this are on a case by case basis.
