Despite years of training and experience — and a vast wealth of medical knowledge at their fingertips — doctors still have difficulty pinpointing certain conditions and diseases. Luckily, the future of artificial intelligence holds the promise of reshaping diagnostic technology and helping doctors finally put a finger on diseases that have a way of eluding detection.
How AI Helps With Diagnoses
Some of the most evasive diseases could be easier to detect with AI-supported diagnostic technology that can scan thousands of medical records and analyze data that might escape the attention of tired, overworked medical professionals. AI can also improve the time-consuming process of classifying medical records.
Electronic health records potentially possess unlocked bits of insight that can lead to robust diagnoses and improved health care. According to Healthcare IT News, 80.5 percent of hospitals have installed at least a basic electronic health records system.
Although patients’ records are largely digitized — easing the transfer of data between medical professionals — most of that information still needs to be read by a doctor. The tediousness of the process — not to mention the role of fatigue and distraction from other pressing issues — can cause a health professional to miss a symptom or misdiagnose a condition.
But as a Stanford dermatologist told The Daily Beast, “‘AI doesn’t get tired.'” That doctor points to how an AI algorithm “learns” to discern subtle patterns it views over large data sets, helping doctors diagnose patients and gain larger insight about diseases.
Making a Difference
Diagnostic technology is helping improve the accuracy of finding tumors. For example, Massachusetts General Hospital researchers utilize a machine learning model for diagnosing tumors. “In this method, researchers fed information on over 600 high-risk lesions—including details about the patient’s age and race, family history, past biopsies, and pathology reports—to an artificial intelligence. In trials, that artificial intelligence was able to interpret that data so effectively that the method resulted in fewer unnecessary surgeries and specifically diagnosed more cancerous lesions,” according to Boston Magazine.
In Washington, D.C., an internist wanted a second opinion on a patient who he believed had rheumatoid arthritis. He posted the patient’s case on an online system called the Human Diagnosis Project, where more than 6,000 doctors can access collective viewpoints to inform medication prescriptions and tests, according to Wired. The system’s natural language processing algorithms mine each case entry for keywords to direct it to specialists who can create a list of likely diagnoses and recommend treatments. The algorithms verify responses to previous cases and weight each specialist’s latest finding, combining them all into one suggested diagnosis.
AI has also captured the interest of doctors in China, where 80,000 radiologists have to annually review 1.4 billion radiology scans. A system that is gaining favor there combines image recognition and deep learning to study healthy X-rays to create a point of reference for pinpointing those with troubling images, potentially reducing their workloads and making a dent in the 600,000 lives lost each year to lung cancer, Forbes reported.
The Future of Artificial Intelligence and Medicine
Medical professionals are quick to note that AI-backed technologies strive to free them from time-consuming, monotonous work and are not meant to supplant the skills and empathy of humans. They hope this diagnostic technology will complement and enhance their work.
Some doctors warn AI could create two classes of medical care, where some patients still seek the care of human professionals while others will accept the diagnosis offered by an algorithm. But as with other industries that are slowly implementing AI technologies, the medical professional will undoubtedly balance time-honored practices that rely on human experience with the best the technology can offer.