Artificial Intelligence and Machine Learning May Help in Understanding MS Cause, Groups Say

Combining data science, artificial intelligence (AI), and machine learning to better identify patterns that may underlie the cause or causes of multiple sclerosis (MS) is the focus of a novel partnership.

Despite numerous advances in MS research and treatments, what causes the disease is still unknown.

“Given the complexity of MS and the urgent need to help patients that are living with this diagnosis, we wanted to explore new ways to infuse technology into our research,” Saud A. Sadiq, MD, director and chief research scientist at the Tisch MS Research Center of New York (Tisch MSRCNY), said in a press release.

Sadiq and fellow researchers at Tisch MSRCNY collaborated with Deloitte, a consulting and advisory services company. Tisch MSRCNY is a nonprofit center specialized in MS, its causes, biomarkers, and other disease research tools.

“We met with Deloitte and discussed the possibility of applying tools like AI and machine learning to narrow down molecules that may be correlated to MS, as well as to help accelerate the discovery process,” Sadiq said.

Using information provided by Tisch MSRCNY researchers, Deloitte identified two different ways to help advance research by applying data science.

First, Deloitte helped Tisch MSRCNY assess markers in patients’ cerebrospinal fluid to identify metabolites (by-products of the different metabolic processes that take place in a cell) associated with MS. The team found molecules potentially correlated to MS within two weeks. According to Deloitte, if this research was done by humans instead of machines, it would have taken up to a decade to finish.

The two then moved their project into a second phase, focused on analyzing B-cells (a type of immune cell) and antibodies.

Deloitte tackled this phase of the project through a type of crowdsourcing approach. In total, 137 teams consisting of more than 400 Deloitte experts competed to develop new analytical models using AI to detect patterns in allele usage (whether the maternally or paternally inherited gene is used), immunoglobulin (antibody) subtypes, B-cell subtypes, genetic edition, and sequence diversity.

This project allowed them to validate a machine learning approach for future MS research.

“In ‘The Age of With,’ a world where humans work side-by-side with machines, AI and machine learning are increasingly being leveraged to solve medical puzzles where human research has encountered challenges,” said Beena Ammanath, AI managing director, Deloitte Consulting.

“Data science is helping organizations find solutions to problems that have yet to be answered through traditional tactics, and I’m so proud that we are working with Tisch MSRCNY to provide talent and tools to help them revolutionize their MS research,” Ammanath added.

Among project discoveries that may be importance was the suggestion of a previously unknown association between plasmablasts (precursor cells of plasma B-cells) and primary progressive MS, Sadiq said.