AI in Healthcare
Perhaps the most exciting space for applying AI to deliver transformative services is in the area of healthcare. Until recently, AI wasn’t a powerful enough tool to undertake the work required. But recent developments, coupled with easy access to AI tools, have resulted in both researchers and start-ups exploring what can be achieved.
As the Ebola epidemic took hold in West Africa in 2014, the healthcare community was under pressure to find a way to bring the illness under control, even if they couldn’t cure it. Typically, the process of simply determining a potential pharmaceutical can take years of research. This is before the drug even makes its way to trials. Thanks to Atomwise and their AI tool AtomNet, existing medicines were scanned to determine which would be a suitable candidate. In the space of a single day, two such drugs were identified, a process that normally takes months or years.
An area where AI often performs better than humans is in the area of data analysis. Due to its skill in pattern recognition, especially of images, AI has been shown to perform as well as the most skilled ophthalmologists in diabetic retinopathy screening. Having been trained on a set of 100,000 images, a deep learning algorithm was able to perform as well as humans using a single image analysis, rather than multiple image analysis used by doctors today. Such an advancement opens up early detection for a broader cross-section of society, especially in countries with limited healthcare provision.
Attempts have also been made to utilize existing electronic health records (EHR) to teach AIs how to detect various illnesses. Unfortunately, the quality of the data is often severely lacking. This could be due to the manner in which it was recorded, with different healthcare providers recording vital parameters in different ways. This has urged various projects to improve the quality of learning data prior to using crowdsourcing. Kaggle is a platform that allows the creation of competitions. The Memorial Sloan Kettering Cancer Center (MSKCC) utilized it to encourage the development of a gene variation tool, based upon their expert-annotated knowledge base.
To sum up, AI will undoubtedly revolutionize the healthcare industry. Robots and smartphone apps will just be the beginning of what patients might notice at first. However, it will be the behind-the-scenes activity into pharmaceutical research, detection and therapy where some of the most exciting AI developments will be found.