This article was reviewed by Jill Hopkins, MD
One healthcare leader is making an investment in personalized healthcare, combining the vast amounts of data being collected during clinical trials with artificial intelligence (AI) algorithms. Roche-Genentech sees a transformation on the horizon, based on the ability to use meaningful data on a large scale.
According to Jill Hopkins, MD, global head, ophthalmology personalized health care, Roche-Genentech, the goal of the company’s personalized healthcare initiative is for patients to be identified at earlier stages of disease, and to be able to predict their most likely treatment needs and responses with a large degree of accuracy. There is also potential to build toward a preventive strategy where vision might be preserved.
The company has targeted ophthalmology, oncology, and neuroscience as core focused therapeutic areas for the development of personalized healthcare solutions.
The company is considering everything from digital tools to predictive and preventive algorithms of disease to change the way patients in these areas are treated. The question is how to take millions of data points being generated, and add machine learning tools for individualized disease progression and therapy.
Roche-Genentech has much legacy clinical trial data to use from the ranibizumab and lampalizumab trials. The database includes more than 13,000 patients, and around 3 million images. It continues to grow through their active phase III programs investigating ranibizumab in the Port Delivery System and faricimab.
“We have worked to optimize the value of the existing data.,” Dr. Hopkins said. “It is clinical trial grade data. It is longitudinal, with multimodal imaging, so it really provides a wonderful initial core data set on which to build some of the machine learning and deep learning proof of concept algorithms.”
1. Arcadu F, Benmansour F, Maunz A, et al. Deep Learning Predicts OCT Measures of Diabetic Macular Thickening From Color Fundus Photographs. Invest Ophthalmol Vis Sci. 2019 Mar 1;60(4):852-857. doi: 10.1167/iovs.18-25634.