Why should doctors be able to understand machine learning?

Artificial intelligence (AI) is becoming more popular across all industries, including healthcare. Innovation has been made possible due to increased data volumes, higher computing power, and new possibilities of machine learning techniques. This is evident in healthcare with a high increase in AI-related researches

Do doctors really need to be able to comprehend machine learning? Here are the top reasons why they have to.

Communication with patients

Machine learning tools are becoming more common in clinics. It is only a matter of time when they become more widely used. Machine learning-based decision support tools will soon be used by your doctor and they will recommend machine learning-based treatments in their everyday practice.

It is important to be able to interpret tests using terms such as specificity, sensitivity, and positive and negative predictive value. A patient with a high PSA does not necessarily mean that they have prostate cancer. We also know that a raised D2-dimer doesn't necessarily indicate a DVT/PE. The tests are sensitive but have low specificity. These are often part of our conversations with patients.

We will need to share more specific terms with patients to discuss AUCs, F1 scores, and other details of machine-learning-based predictions.

Many patients will also want to know how these new interpretations and predictions work. Patients will be less satisfied with the explanation that "then a very complex algorithm spits off a prediction", than with "the algorithm has been trained on 1000s patients to spot patterns that predict X and Y, so we can take into consideration your age, genetic profile, and comorbidities to give you a personalized prediction." Probably you will have to explain to them what is present in brain datasets.

Contribute towards research and innovation

Artificial intelligence in healthcare is one of the most exciting and dynamic research areas. There are so many possibilities and even more things to explore. As a doctor with a good understanding of ML, there are many ways you can help this field.

Large amounts of data are stored in hospitals, which can be used to create exciting AI research in healthcare. The data is often left unexploited if there are not enough people within the department to understand how it can be used. Also, you could assist in establishing a new department for cutting-edge AI research if you are an expert in ML.

Collaboration with many companies in the healthcare AI sector is also possible. They often include a mix of researchers, machine learning engineers, and healthcare professionals. Although ML engineers will be building the tools, healthcare professionals will help them. It is in high demand to be able to communicate with patients and translate their medical information into 'data' or 'variables'. It is possible to provide insights into what can be incorporated into the medical workflow, and what might be of limited use. Is it possible to train an algorithm that can recognize absolute observations/vital signs or fluctuations relative to the individual's baseline? A high-sensitivity algorithm for AKI that sends out pop-up notifications would be either beneficial or detrimental. Working within healthcare systems provides many insights. We shouldn't let those who don't have the experience to develop healthcare AI forget about them.

Participate in the discussion

Understanding ML will allow you to cut through the hype, and help shift the conversation towards a more measured and accurate assessment of the current situation. You will be able to tell why AI is better at diagnosing pneumonia than radiologists if a top machine learning researcher says so.

One of the most important transformations in our generation will be the implementation of ML throughout society and healthcare. We, doctors, want to be involved in how this happens. There are many ethical issues that must be addressed, and they do not always match the financial incentives offered by companies in this area. Understanding ML will allow us to become informed, responsible members of the medical community and contribute to policy and discussion, thereby facilitating AI's safe, effective, and patient-centered introduction.

How?

Understanding machine learning is the next step. Our advice is to check for recommended resources like https://www.medicaldata.cloud/ to understand what the healthcare database is and then to maintain current knowledge.

(Devdiscourse's journalists were not involved in the production of this article. The facts and opinions appearing in the article do not reflect the views of Devdiscourse and Devdiscourse does not claim any responsibility for the same.)