Good medicine meets artificial intelligence PDF Print E-mail
Saturday, 02 December 2017 04:18
AddThis Social Bookmark Button

Sarthak edit

New cures for scourges like cancer and HIV, and machine learning to make traditional diagnostics smarter

Technological and scientific breakthroughs are outlining a future that could look vastly different from today, and help cure many of the problems the world is facing—from cures for dreaded diseases to making diagnostics cheaper and more easily accessible—at a much faster pace than ever before. In the case of malaria, for instance, as compared to the traditional treatment via drugs—increased drug resistance puts a brake on the efficacy of most cures—Indian scientists have successfully indicated the treatment can be shifted to using proteins to prevent the pathogen from infecting RBCs; in 2015, 212 million people were affected by malaria, and around half a million died. In the case of HIV that affected over 1.5 million new people every year between 2010 and now—resulting in a stock of around 37 million by 2016—a new university/research institution-government-pharmaco collaboration in the US has developed a super-antibody that can neutralise 99% of the HIV strains that exist today.

And in the case of cancer—expected to rise from 14.1 million new cases in 2012 to 21.7 million by 2030—the treatment could soon move away from the traditional chemo- and radio-therapy to engineered immune-cells therapy; this started 5-6 years ago, but advances now are far more rapid than in the past. While it was originally developed by the Children’s Hospital of Philadelphia, many private firms and public institutions are trying to perfect this. In this cell therapy, called chimeric antigen receptor (CAR) T-cell therapy, doctors harvest a patient’s T-cells which, as part of the immune system, fight against infection. These harvested T-cells are then reprogrammed to target cancerous cells in the patient’s body. CAR-T has had a 93% success rate in trials with patients with advanced leukaemia.But there have been instances of unpalatable side-effects also. Juno Therapeutics, an American company working on a CAR-T solution for cancer, called off the trial of JCAR015, the immunotherapy drug it had developed, after the US FDA flagged patient-deaths from side-effects. However, CAR-T based therapeutics developed separately by Novartis and Kite have got the FDA nod; the University of Pennsylvania (UPenn) is also conducting its own trial, one of the 800-plus studies in gene and cell therapy currently underway. While commercial manufacture is expensive, a Canada-based collaboration between GE, the government of Canada and a non-profit funded by both private and public players, including leading Canadian universities, is trying to make the therapy scalable and affordable in the long run.

And, in the case of diagnostics, as an example of the work being done by AI, a team at GE Healthcare has designed an app that relies on machine learning to make X-ray imaging at hospitals more efficient—X-rays comprise more than half of all medical scans, and within X-rays, chest X-rays account for half. Errors in X-ray imaging impose a double-penalty—the patient is exposed to more radiation, and the hospital/medical facility must deal with increased scanning burden that slows the flow. GE’s app automatically pulls all of the information about imaging errors and repeat-imaging into a one-page dashboard for easy review. The first iteration of the app will help hospitals reduce patient exposure to radiation and make radiology departments more efficient. In the future, they may even assist in diagnosis; GE cat-scanners already contain data on lakhs of previous scans to help radiologists better their diagnosis. Though the host of diagnostic apps using machine learning are yet to get regulatory approval, the hope they hold out for quicker, more reliable and more economic diagnostics are tremendous.


You are here  : Home Miscellaneous Good medicine meets artificial intelligence