Tackling Alzheimer’s Biomarkers Using Machine Learning: Interview with IQuity CEO, Dr. Chase Spurlock

By | November 17, 2018

Losing one’s memories, and therefore one’s identity, must be a terrifying experience. With a global estimate of almost 50 million people living with dementia, predicted to rise to more than 130 million by 2050, the burden of Alzheimer’s and dementia are indisputable. While there are still no approved drugs for the treatment of Alzheimer’s, there is an urgency for developing diagnostic tests and the identification of biomarkers that enable the early detection of the disease. In light of this need, diagnostic technology company IQuity (pronounced I-Q-witty) has recently received a grant from the National Institutes of Health (NIH) to build a platform around the early detection of Alzheimer’s. We sat down with IQuity’s Founder and CEO, Dr. Chase Spurlock, to talk about this effort and the company’s progress with other diseases they have tackled.

Mohammad Saleh, Medgadget: Tell us about yourself and how you came to be a part of this work.

Chase Spurlock, IQuity:  I’m the CEO of IQuity, a tech company based in Nashville using data science to detect and monitor disease. At IQuity, we can use a variety of data sources – ranging from blood samples to large, population-level healthcare datasets including insurance claims or electronic health records to identify disease and help optimize patient journeys. A large portion of my work has focused on autoimmune diseases through research initiated at IQuity and through my faculty position at Vanderbilt. Autoimmune diseases affect you and someone you know – these are conditions like multiple sclerosis, rheumatoid arthritis, Crohn’s disease and others where the immune system harms the body’s tissues and organs. We can’t cure these diseases, but we can treat them. If we can detect the disease early, we have the best chance for optimal long-term outcomes including reduced disability and longer life expectancy.

The first demonstration of our use of data science to build tools to help providers diagnose and treat these illnesses was to create blood-based tools that doctors can use to make fast, accurate diagnoses. We created a platform that looks at RNA data, a real-time snapshot of what’s taking place inside the patient’s blood cells, and then builds computer models to allow us to examine new patient samples and ask if they look like patients with specific diseases we may have seen before. To date, we have commercialized three tests leveraging our technology for multiple sclerosis, IBS (irritable bowel syndrome) and IBD (Crohn’s disease and ulcerative colitis), and fibromyalgia syndrome.

Apart from our blood-based studies, we began using our disease knowledge and data science expertise to look at large, population-level datasets to see if we could craft similar insights that enable early disease detection and patient monitoring. Layering claims data with datasets that help us understand the social determinants of health, we have been able to mirror our blood testing work and create a technology that enables faster disease detection, correction of misdiagnosis and the ability to monitor patient disease at scale. Monitoring is critical to ensure that treatment plans remain effective and prevent adverse events. Our approach makes it possible to alert care teams so that interventions can occur faster than is conventionally possible today.

As we look to the future and the broad applicability of our technology, we began exploring whether we could use our blood-based approach or population analytics platform to tackle new disease indications. Alzheimer’s disease quickly rose to the top of the list because this fatal neurodegenerative disease affects over five million Americans and often goes unnoticed until symptoms manifest 15-20 years after the damage to the brain starts. We feel that our approach using blood-based technology for population-level analytics holds extraordinary potential to make an indelible mark on how clinicians can approach Alzheimer’s care in the years ahead.

Medgadget: Can you give our readers a quick overview of what the current status of Alzheimer’s early detection and drug research is?

Spurlock: We need tools that can accelerate the diagnosis of Alzheimer’s disease – whether these innovations come in the form of biomarker development, digital technologies, or a combination of the two. Both approaches will allow us to understand how the disease develops and progresses, the patient populations affected by Alzheimer’s and provide opportunities to refine future clinical trials by enhancing patient recruitment efforts. There are no approved treatments that can slow the progression of Alzheimer’s disease – they are primarily used to treat symptoms and are useful for only a short window of time. In the future, approved drugs for Alzheimer’s will be most effective when administered early. The diagnostic tools we have proposed to develop through our NIH-funded studies seek to identify patients at the earliest stages of the disease and could potentially be used in clinical trials to assess treatment responses. If we can identify patients early and include them in clinical trials, we will have a better chance of finding therapeutics that can alter the trajectory for this devastating condition.

Medgadget: What makes IQuity uniquely adept at​ tackling this problem?

Spurlock: At IQuity, we have successfully commercialized tools that can be used to identify disease. Using multiple sclerosis (MS) as an example, we have shown that we can identify patterns in blood at the time of a clinically isolated syndrome, which is the major clinical precursor to MS. In MS and fibromyalgia, have shown us that these long non-coding RNAs (lncRNAs) exhibit high levels of disease-type specificity meaning that lncRNA patterns measured in blood are often specific for one disease and can be used to differentiate among similar conditions and healthy individuals. The preliminary studies that we outlined in our recent Alzheimer’s grant highlight the utility of these biomarkers to distinguish Alzheimer’s disease from other neurological illnesses.

As an expansion of our work in genomics, we are eager to explore the use of the longitudinal health records in our population datasets (insurance claims data or electronic health records) to assess whether our existing data science approaches could be applied to identify early patterns for Alzheimer’s. These efforts will help us understand environmental, racial or other social factors that can explain how this disease may arise and could lead to opportunities for early intervention.

Across both areas of our business, I believe IQuity is uniquely poised to make discoveries that can improve diagnostic tools, inform pharmaceutical development and enhance the quality of life for patients living with many diseases. While autoimmune conditions have traditionally been our focus, the opportunities to help patients with a variety of medical conditions, including Alzheimer’s, are possible with the infrastructure we have put in place.

Medgadget: Can you tell us about​ the technology driving the Alzheimer’s detection efforts? How does it work and where does machine learning factor in?

Spurlock: The funding we recently received will support our work to identify and measure lncRNA found in blood and build software capable of detecting specific RNA patterns for Alzheimer’s disease. RNA-sequencing will allow us to identify candidate RNAs we will then validate in a larger cohort of Alzheimer’s patients, disease controls and age-matched healthy controls. Data derived from this expanded cohort will then be used to train machine learning models to determine if these biomarkers can be used to construct a specific blood-based pattern for Alzheimer’s. Independent blood samples not included in the training will be used to test if the model can accurately and reproducibly distinguish among these groups. This work will consist of patients from across the United States, allowing us also to examine if geography can influence the expression of the RNAs we identify.

What’s interesting about this approach is that machine learning helps us identify the RNAs present in a blood sample that hold the highest level of ‘predictive power’ and determine among those we select which RNAs are more or less important in determining the presence of Alzheimer’s or any other disease we choose to study. While we are explicitly using our approach to build diagnostic strategies to identify presence or absence of disease, use of these methods could uncover novel targets for therapeutic intervention if these biomarkers are found to change as a consequence of intervention.

Medgadget: Can you maybe share some specific success stories? How accurate has the platform been so far?

Spurlock: Pilot funding from the NIH will be used to launch our Alzheimer’s work. It will allow us to expand our focus beyond the company’s traditional emphasis on autoimmune disease to test if our approach can be used to detect Alzheimer’s accurately. What we have documented thus far is that our blood tests are more than 90 percent accurate for the autoimmune and related conditions that we’ve examined.

Medgadget: Do you think work like IQuity’s early detection efforts for Alzheimer’s can lead to a renewed interest in searching for therapeutics for Alzheimer’s?

Spurlock: Absolutely. If we can identify these diseases early using genomics or population-level analytics, we will have a series of useful tools that can help determine the right patients for clinical trials. Identifying these patients at the earliest moment means they can take an investigational drug and have the best chance to reap the benefit of these new therapeutics. As Bill Gates and the Alzheimer’s Drug Discovery Foundation have correctly stated many times, the new drugs currently under development will be most successful in slowing early Alzheimer’s. All too often, patients who develop Alzheimer’s are caught many years after the disease has already caused damage to the brain.

If we can identify diagnostic biomarkers, we can better understand the genetics of how these diseases manifest and begin to understand, at a molecular level, the mechanistic underpinnings of the disease and create targeted assays that can look for these changes. Using population-level analytics to ask these same types of questions, we can begin pinpointing the social determinants of disease and from these studies further refine the inclusion and exclusion criteria for clinical trials. The combination of molecular and population analytics, I believe, will enable us to understand better the types of patients who ultimately arrive at an Alzheimer’s diagnosis and create a window for early intervention.


Medgadget: Where do you see IQuity in the next 5-10 years with your work in Alzheimer’s and other areas?

Spurlock:  IQuity’s future is very bright. We’re at a significant inflection point in healthcare where we are starting to see artificial intelligence/machine learning tools implemented more often for clinical use. At IQuity, we are creating a series of platform technologies that we can apply to many areas of human health and disease. While we have focused on autoimmunity and related disorders, we have found that our blood-based and population-level analytics can affect many conditions. The work we’ve done has shown that we can predict oncoming diagnoses, spotlight misdiagnoses, and stratify severity of disease. A unique aspect of our approach is the ability to trigger not only an alert, but embed a context or reasoning into each alert that helps those receiving this information understand how our tools arrived at a result. I think this level of insight is going to be at the heart of how we can positively shape healthcare to make the patient experience better, improve outcomes and lower healthcare costs.

Medgadget: When should we expect to be hearing more data about IQuity’s Alzheimer’s efforts?

Spurlock:  The focus of this project hits close to home for many on our team. My own family has witnessed firsthand the negative consequences of early Alzheimer’s and dementia. We often wondered if we could have done something differently to recognize or treat the underlying causes of the disease – but it just wasn’t possible. There remains so much work to be done to make this a reality. At IQuity, the second we received funding we immediately began work. Our laboratory efforts are underway to refine the biomarkers we have already identified, and I would expect data to emerge in the next several months that we intend to share at conferences or in peer-reviewed journals.

For more information, check out IQuity’s website or read our previous interview.