We recently announced a computational Challenge, the ALS Prediction Prize Challenge, with Prize4Life to better predict the progression of disease in ALS patients. Having a better understanding of how this deadly neurological disease, also known as Lou Gehrig’s disease, progresses could lead to breakthroughs in treatment and quality of life for patients. We recently spoke with Dr. Melanie Leitner, Chief Scientific Officer for Prize4Life, about the Challenge.
Hello Dr. Leitner. First of all, can you tell us a bit about ALS, who it typically strikes, and its impact on patients?
Amyotrophic Lateral Sclerosis, also known as Lou Gehrig’s disease in the US and Motor Neuron Disease in other parts of the world, is a progressive fatal neurodegenerative illness that attacks motor neurons. When motor neurons die, the ability of the brain to control muscle movement is lost, leading to paralysis. Unable to function, muscles atrophy. Eventually, all muscles under voluntary control are affected and patients in time lose their ability to walk, talk, swallow, and breathe. During this terrible process, the mind typically remains intact, observing the loss of each function. When the diaphragm and chest muscles fail, patients stop breathing on their own. The majority of people with ALS die of respiratory failure within 2 to 5 years of noticing the symptoms.
The ALS Association estimates that there are about 30,000 people living in the US today with ALS. Ironically, ALS strikes as many or more people than some other better known neurodegenerative diseases, such as Huntington’s Disease or Multiple Sclerosis, but because ALS is so rapidly fatal, at any given time there are fewer people living with ALS.
People of all races and ethnic backgrounds are susceptible to ALS, which typically strikes people between the ages of 40 and 70, but it is known to affect younger and older people as well. The disease occurs slightly more often in men than women.
This Challenge is focused on developing accurate and predictive indicators for disease progression. Why is this so important?
The question of accurately predicting disease progression is important for two key reasons. First, as an ALS patient or caregiver of a patient, it is frustrating that, unlike other diseases, when you’re diagnosed with ALS there is currently no way of knowing at the outset if you’re going to be like Lou Gehrig and have a very rapid course to your disease or if you’re going to be one of the rare individuals like Stephen Hawking who live more than 10 years with the disease.
Second, being able to accurately predict disease progression would make it easier to design and conduct effective ALS clinical trials. Right now, because researchers can’t tell whether a given ALS patient will live for two, five, or ten years, clinical trials require observing a large number of patients over a long period of time to determine whether or not a potential new therapy is effective.
Are there no current predictors of ALS disease progression?
There is currently a widely used rating scale: the ALS Functional Rating Scale (ALSFRS) which is a good measure of a given patient’s current ALS status and over long periods of time is a relatively good predictor of ALS disease progression. But ALSFRS is not predictive in the short term and is mostly useful at the population level, rather than for any given individual, due to variability inherent in the measure.
As a computational Challenge, Solvers will have access to a set of validation data. Tell us more about the Challenge dataset, as well as the larger database that Prize4Life is building.
We wanted to provide Solvers with a rich set of data on which to build and test their algorithms. The PRO-ACT (Pooled Resource Open-Access ALS Clinical Trials) database will be the largest open-access ALS clinical trials database ever built. Merging the data from large, completed, privately and publicly funded ALS clinical trials will enable analyses that heretofore had been limited by the small numbers of patient data points available. Upon completion, PRO-ACT will contain more than 7,500 complete clinical records, and this Challenge uses a subset of nearly 1,900 clinical data records from the total PRO-ACT database.
Each fully de-identified record includes demographic, family and medical history data, and drug status (placebo vs. medication) and spans 11-12 months of multiple measures of clinical data including blood chemistry, hematology, urinalysis, vital signs, and functional rating scores and sub-scores. This subset of the full PRO-ACT dataset being used in this Challenge contains over 5 million data points.
Participating Solvers will also be able to submit iterative versions of their solution for validation and feedback. In fact, Solvers can do this up to 100 times. What is the advantage to Solvers to participate in this process?
You were probably expecting me to say that participating in the validation phase is a way for Solvers to test their predictive algorithms as they develop them, get real-time feedback, and compare their relative rankings against their fellow Solvers. But Solvers should take advantage of the leaderboard because it will be fun (and they could be the envy of fellow Solvers around the world)! From our perspective, we also encourage Solvers to participate because it will help ensure that the winning solution at the end is as robust and useful as possible.
Prize4Life has employed Challenges to solve other problems relative to ALS. How does this crowdsourcing approach fit within the mission and strategy of the Prize4Life organization?
Our mission is to accelerate the discovery of treatments and a cure for ALS by using powerful incentives to attract new people and drive innovation. We were founded on the belief that the next ALS breakthrough could come from anyone, anywhere, at any time. Crowdsourcing is therefore a key pillar of Prize4Life’s strategy and we believe it is one of the most effective forms of innovation.
Dr. Leitner, thanks very much for speaking with us.
Thank you, and thanks in advance to all Solvers participating in this important Challenge.