Is AI Better at Diagnosing Disease than a Doctor?

The use of AI in medicine has moved beyond the realm of theory. Artificial intelligence is changing healthcare right now.

Artificial intelligence has been increasingly used in many disparate fields including robotics, finance, and air traffic control. But one of the most promising uses of AI is in medicine.
This is an attractive concept because a human doctor cannot recall all the information necessary to make an accurate diagnosis at a given moment. Considering the huge amount of research papers and textbooks available, we cannot expect a doctor to master every aspect of medical care and recall every detail that they have learned.

Precisely for this reason, AIs may prove to be a great boon for healthcare when we need to examine a great deal of information. Certainly when it comes to assessing large volumes of data, AIs excel far beyond what human physicians are capable of. But can they use this ability to diagnose and even treat patients better, as well?

Let's look at three cases of how AI is currently being used in the medical field, all with relatively similar concepts:

1. IBM’s Watson used to diagnose cancer
2. Google DeepMind's AI used to make accurate eye diagnoses
3. Babylon’s AI-based app used to suggest the course of medical care

IBM’s Watson Cancer Diagnosis

Watson, a question-and-answer computing system, already has a great resume—in 2011, it won a $1 million prize on Jeopardy!.

Now Watson is giving the medical world its best shot. This August, doctors at a hospital in Japan misidentified a 60-year-old woman’s leukemia. But Watson examined a vast database of 20 million research papers and make a successful diagnosis—in only 10 minutes.

Doctors need to master a large volume of research results, medical records, and clinical trials. This is not easy at all, given the limited capacity of a human brain. On the other hand, an AI-based system can be utilized to prune out the irrelevant data and help the doctor think more clearly focusing on the vital data.

Image courtesy of IBM.

Watson analyzes the medical records and notes of a patient which is in plain English and gathers the patient’s key information. This information, along with the data obtained from a wide variety of sources (including 12 million pages of text, nearly 300 medical journals, and hundreds of textbooks) is utilized to select some treatment options for that particular patient.
Moreover, Watson provides supporting evidence for each treatment option as well as the warnings and toxicities for the drugs it proposes prescribing. By ranking the treatment options and linking them to the relevant supporting evidence, Watson helps doctors successfully make the important decisions for the patient.
Watson and other AI systems can be successful in cases with large comparable data. In such cases, it is possible to plug the right data into the AI-based system and let the computer sift through the database—instead of relying on the much more limited knowledge of a human brain.

DeepMind’s AI and Eye Diagnosis

Google’s AI division, DeepMind, is planning to make its technology available for the Moorfields Eye Hospital in London. This cooperation will utilize DeepMind’s AI technology to analyze more than one million eye scans and find out the early warning signs of visual degeneration. Machine learning is expected to recognize those symptoms that even experienced doctors may miss.

The project, which is DeepMind’s first attempt to use AI for a medical purpose, mainly aims to find the symptoms of visual problems caused by diabetes and age-related sight degeneration. These two cases are the most important causes of sight loss in the U.K and can be effectively treated provided the early detection of the disease is possible.

Diabetic sight loss is one of the most widespread and severe kinds of blindness around the globe and Mustafa Suleyman, co-founder of DeepMind, hopes that early detection of the disease will enable doctors to prevent 98% of this kind of blindness in the world. Therefore, there is much at stake.

Interestingly, the idea of this project occurred to Pearse Keane, a doctor at Moorfields, when he noticed that DeepMind’s AI is capable of teaching itself to play Atari games. Then he asked DeepMind to use its machine learning for this project.

Babylon’s Artificially Intelligent Doctor

Babylon, a U.K.-based health service, raised $25 million early this year to develop an AI-driven app for healthcare services. Babylon currently provides video consultations with doctors. Future versions, when ready for a full commercial launch, will act as an artificially intelligent doctor. The AI-driven app will receive the patient’s symptoms and check them against its database. Babylon will use a database similar to that of Watson.

Images courtesy of babylon.

In addition to its database of papers and textbooks, Babylon will consider the individualized history and circumstances of the patient including family health history, medical records, daily habits, heart rate, cholesterol levels, allergies, and more. Some of this data will be obtained by constantly monitoring the readings of the wearables. Examining hundreds of millions of combinations of symptoms, Babylon will offer the appropriate pieces of advice.

Ali Parsa, the founder of Babylon, aims to create a health service that everyone can afford. He expects that the app will be able to even predict the illness before it occurs. As an example, if a patient's heart rate increases without involving in any physical activity, it is likely that they are either going through stressful experiences or are dehydrated. Monitoring these changes, Babylon will be able to offer the best course of action for each individual.

---

Incorporating AI into the medical field has the potential to change—and possibly vastly improve—healthcare in its entirety. From improved diagnostic accuracy to better-optimized treatment plans, AI could be the key to better medical care for doctors and patients alike.

While it's possible that AIs could replace human doctors in certain areas in the future, for now they are proving to be a valuable tool for doctors to use in tandem with their own expertise.