OR WAIT null SECS
While at the 2014 Bio IT World conference, I had the opportunity to interview Pek Lum, Ayasdi's Chief Data Scientist.
While at the 2014 Bio IT World conference, I had the opportunity to interview Pek Lum, Ayasdi’s Chief Data Scientist. Ayasdi is an analytics platform which enables researchers to understand their data through topological networks.
Moe Alsumidaie: How Does Ayasdi envision the future of medicine?
Pek Lum: At Ayasdi, we let the data speak for itself, for example, we are able to group patients through targeted visualizations, such as biomarker electronic medical record, clinical, and demographic data, and we optimize the shape of these visualizations to interrogate the data with statistical analysis. This allows the medical community to optimize therapeutic administration through visualization by segmenting targeted patients in a better way. To elaborate, we provide a more detailed view of biomarkers, the impact of genetics on therapeutic outcomes, and adverse event discovery.
MA: Ayasdi did a few case studies with Mount Sinai and the Michael J. Fox Foundation. Can you elaborate on the results?
PL: At Mount Sinai, we did a collaborative with Joe Dudley, where we accessed a rich database of medical records, including diagnoses, CPT codes, bloodwork, and genotype data of about 30,000 patients. We then specifically focused on over 2000 patients diagnosed with diabetes. We first evaluated medical records to uncover trends in medical outcomes, for instance, whether a patient is reacting to a particular therapy, and we then linked these trends with patients’ biobank data to understand whether specific genotypes had an impact on patient diagnoses and therapeutic outcomes. Through aggregated genotype visualization analysis, we discovered that there were at least three subgroups associated with diabetic patients, which is now changing the way we treat and diagnose diabetic patients.
At the Michael J. Fox Foundation, we analyzed data derived from smart phones that were carried around by Parkinson’s disease patients as well as a group of non-affected controls for about a month. Data was collected every second from the phones. In addition to being able to separate controls from the patients, we also uncovered two subgroups of Parkinson’s patients. We are not sure what those two subgroups are clinically because that information was not available but these two groups differ in their movements at certain frequencies. With this kinds of analysis, we hope to be able to assist physicians with understanding patient disease progression by how the patient moves on this visualization or “map” after being treated, and correspondingly, allow physicians to optimize their treatment regimens for Parkinson’s disease patients in real time.
MA: How will Ayasdi’s platform help in therapeutic targeting and clinical trial design?
PL: Ayasdi’s platform can improve drug development in several ways. Firstly, enhancing patient targeting through biomarker analysis. Sponsors tend to express concerns that targeting a medical product may reduce the applicability of the product on the patient population after it has been approved; on the contrary, targeting improves the success of a medical product’s commercialization potential because sponsors can demonstrate high effectiveness outcomes, which means insurance companies will be more compelled to cover the medical product, and sponsors can better balance the pricing of their medical products due to uniqueness. Herceptin is a great example of a successful targeted therapy. Secondly, the Ayasdi system can reduce the amount of patients that are needed in a clinical trial because sponsors can tailor a clinical trial for subjects best predicted to respond through biomarker screening, which brings us to my last point, the Ayasdi analysis and visualization system can significantly improve therapeutic outcomes in clinical trials.