It is no secret that artificial intelligence (AI) is making huge strides in the healthcare and health tech industries. Patients and providers alike look forward to the positive effects it could have on electronic health records, the patient care model, and the revenue cycle. As the technology progresses, however, scientists and clinicians are finding that AI may be capable of so much more.
The latest news in AI has many pondering its diagnostic possibilities. The release of the Apple Watch Series 4, which can identify atrial fibrillation, and the purported illness-detecting capabilities of Amazon’s Alexa bring to light an entirely different yet exciting field of opportunities. Several technology firms have identified and embraced this potential, spearheading artificial intelligence projects that aim to improve diagnosis and detection for specific conditions. Let’s take a look at a few of them.
Combining artificial intelligence with PET scans resulted in an algorithm that has shown remarkable success rates in the early detection of Alzheimer’s Disease. In fact, in preliminary studies, the AI-based algorithm was able to accurately predict which patients would develop Alzheimer’s in over 90% of cases. The case studies were relatively small and the machine learning is still being fine-tuned but if the algorithm continues to perform at this level, then providers may be able to begin intervention as early as six years prior to an official diagnosis.
A recent AI-based algorithm has shown measurable success when asked to detect potentially dangerous cervical lesions from an image. The algorithm is known as the Automated Visual Evaluation (AVE) algorithm and has been proven to be more reliable than traditional cytology tests in preliminary studies. To date, the technology has been validated by the National Cancer Institute and the National Library of Medicine.
Similar to the AVE algorithm, the new IDx-DR technology uses AI to assess an image of the retina. The program uses both artificial intelligence and cloud computing in an attempt to identify signs of Diabetic Retinopathy. The process takes approximately 20 seconds, with the result rendered in the form of a follow-up recommendation. The IDx-DR recommends an immediate ophthalmology appointment if signs are present, or a re-screen in one year if signs are absent.
DeepGestalt, an algorithm developed by genomics company FDNA, has demonstrated at least a 90% success rate when asked to identify patients exhibiting phenotypes consistent with certain genetic disorders like Cornelia De Lange syndrome and Angelman syndrome. The algorithm divides the facial image into separate sections and analyzes each of them for consistency with known genetic disorders. It then reassembles the image and makes a decision based on the composite result.
The clinical use of artificial intelligence in diagnostics will continue to develop as the technology behind AI improves and we gain more experience with it. While these advancements are exciting and innovative, it is important to keep in mind that they are not yet a substitute for current diagnostic practices, but rather a complement to them in an effort to increase diagnostic accuracy.