Imagine waking up to excruciating pain in either your face or neck. The discomfort can vary in intensity, but symptoms like nausea and sensitivity to light and sound can impair mobility and basic functioning.
This is what it could feel like to have migraines, a debilitating condition that, currently, has no cure.
However, there are some promising treatments in the pipeline. Pharmaceutical companies like Eli Lilly, Amgen and Allergan are racing to develop monoclonal antibody treatments targeting a protein called the calcitonin gene-related peptide (CGRP). This small protein can widen blood vessels and helps transmit pain signals throughout the body, according to The Verge. These experimental drugs are being designed to essentially block the function of CGRP.
Right now, preventative measures are the most effective way for patients to manage migraines. One company is offering a piece of technology that does just that, by alerting patients as to when they are most likely to experience a migraine episode.
The mobile application is called Migraine Alert, and was released by Second Opinion Health, a company based in Mountain View, California.
The company’s chief executive officer is Simon Bloch, who previously worked at Samsung Electronics on mobile technologies and wearables. It was there he conceived the idea of using these tools to assist people with migraines in conjunction with current chief technology officer Jitendra Kulkarni.
“Basically, what we do as a company is use artificial intelligence and machine learning technologies to help people manage their chronic diseases,” said Bloch in an interview with R&D Magazine.
The app is intended for patients diagnosed with intermittent migraines, which is defined as those who experience four to 14 episodes a month.
Migraine Alert works by using machine learning to collect and analyze triggers that correlate with the onset of an episode like weather, stress, activity, and sleep.
“Migraine is a neurological disease and there are no biomarkers you can spot,” continued Bloch.
He elaborated that when his team thought about identifying triggers for the condition,they classified them into two groups, controllable and uncontrollable, since a combination of these triggers affect every individual in a different way.
Weather, stress, activity, and sleep are the four major triggers that cover about 75 percent of the likely reasons involved with causing these attacks, according to Bloch.
A clinical trial performed by the Mayo Clinic generated positive feedback for the initial program.
“We gave participants a Fitbit, had them pair it up with a smartphone, and then continue to live their life as they had before enrolling in the trial,” elaborated Bloch.
Participants were asked to submit information through the app when they experienced a migraine. The things they would need to add include a start time, end time, severity levels, and other factors they may have been exposed to either before or during the migraine.
An additional step involved the machine learning algorithm working in the background to “label” this incoming information to add more context to the analysis, like tracking the time frame of before, during, and after they had a migraine. Weather data gets obtained through the closest weather station in proximity to the user. Data about activity, sleep, and heart rate were collected through the Fitbit.
“We grade all that information, we store it in a database, and then we continuously monitor the machine learning algorithms to see when the triggers begin to correlate,” elaborated Bloch.
The measurement used in this scenario was called the Area Under Curves (AUC) to help build a strong predictive model.
“When the …[correlation of triggers]…start happening, we created an individual prediction model. Next, we start showing the individual their prediction number of a probability of having a migraine attack which is presented in terms of forecast,” said Bloch.
“It’s very similar to a weather forecast. We say, ‘The forecast for you getting a migraine attack for 11 a.m. today is so and so percent.’”
Higher than 66 percent indicates the person would need to follow his physician’s instructions and take possible actions to prevent an attack. A low percentage would mean that the person is experiencing symptoms of a headache and could take painkillers to treat the pain.
Overall, this program can generate very accurate results. By logging fifteen migraine episodes, the application delivers accurate results of an attack by 85 percent, which is higher accuracy than what was previously thought to be possible.
Potential for partnerships
Second Health released the first commercial iteration of its platform in early August 2017 for the iPhone, but there’s still more work that needs to be done to keep refining the program.
Currently, the company is in the middle of a second clinical trial being conducted at the University of Southern California. The trial will last for about five months. The investigation will be used as a way to learn more about the migraines of an individual and validate the accuracy of the predictive model in a blind study.
Bloch ultimately envisions this program as being a companion to managing this chronic condition, especially with pharmaceutical companies and health insurers.
Teva Pharmaceutical and Amgen have filed biologics applications for their respective migraine drugs, but there’s still no infrastructure in place for how health insurers will reimburse different parties for these drugs.
“We can help the drug companies increase the perceived efficacy of the drug and work with insurers to better identify certain people that they will be willing to reimburse for the cost of using the drug,” Bloch continued.
Finally, Bloch considers this program to be a form of proactive technology that provides consumers with a higher level of self-care compared to what was used in the past.