AI drives new signal-detection efforts in Alzheimer’s and ALS.
The life sciences industry continues to embrace artificial intelligence (AI), algorithms, and machine-based learning in its pursuits to collect and analyze important pharmaceutical and healthcare data. I recently had two conversations with people using these tools to improve the ways that Alzheimer’s disease and amyotrophic lateral sclerosis (ALS), respectively, are diagnosed and, ultimately, improve patients’ chances of mitigating the symptoms and living longer.
David Bates, PhD, co-founder and CEO of Linus Health, discussed work his company is doing to boost the Alzheimer’s detection process. AI and machine learning (ML) are at the core of his technology, which focuses on identifying and recognizing digital biomarkers for the disease.
“We built a digital platform that can deconstruct some human behavior into discrete signals,” he says. “We have integrated what’s known as the Boston Process Approach, which was developed several decades ago. The process involves observing the subject’s process of completing the test. That provides as much information as the results of the test itself.”
Unlike traditional diagnosis methods, digital biomarkers don’t rely on measuring how the patient feels. Instead, they track physiological and behavioral data points.
According to Bates, the algorithms require a large number of data points before the platform can start recognizing trends that signify that something might be wrong.
This is especially important for Alzheimer’s diagnosis. As Bates explains, by the time a patient is showing symptoms, the disease has likely progressed for about a decade. Patients may notice memory loss or disorientation, but by that point, they’ve already suffered from a loss of neurons, which can’t be recovered.
Obviously, using digital biomarkers for early detection of disease is helpful for patients. It may enable them to start taking medication or adjust their lifestyle to slow down or stave off the effects of the disease. There’s another important benefit for integrating these biomarkers, however. Testing new therapies for Alzheimer’s has long been an arduous and expensive process, due to the difficulty of diagnosing the condition and finding proper candidates for the testing and control groups in clinical trials. Identifying digital biomarkers can make it easier for researchers to identify and target patients who may have the disease but haven’t started showing noticeable symptoms yet.
“We take over 100 biomarkers in a test alone, and it’s an assembly of them that lead you to conclude that this person is likely suffering from Alzheimer’s disease,” says Bates. “It’s been shown that you can differentiate between someone who has Alzheimer’s and someone who has vascular dementia, or Parkinson’s[, for example]. “The relative deviation from norm of those biomarkers give insight into not only if they are impaired, but also what kind of impairment they’re suffering from. This is important for clinical trials because it helps you reduce the heterogeneity of the population, so you’re not lumping everyone into one bucket.”
Indu Navar also spoke with Pharm Exec about her work toward improving the diagnosis process. With a technology background and holding master’s degrees in computer science and electrical engineering, Navar was on the founding team of what is today known as WebMD. She also founded Everything ALS, a non-profit that uses AI, remote patient monitoring, and propriety algorithms to identify key markers in diagnosis. Navar’s husband tragically passed away from ALS in 2019. He was 49 when he showed the first symptom, foot drop, but it took two-and-a-half years to be formally diagnosed, Navar says.
“The goal is to bring technology innovations to ALS, and if we crack ALS, these methods would be applicable to other neurological diseases,” she says. “We keep thinking that we understand more science and biology, but we all know that if you don’t know how to measure it properly, it’s really hard to prove something. Again and again, ALS patients are put into trials who are in the equivalent stages of the disease to Stage 4 cancer and hope that a miracle will happen.”
According to Navar, one of the goals in ALS is to move away from self-reported data, which can be inaccurate due to patients not understanding the scales properly. Digital biomarkers, however, provide consistent and accurate data points that can be measured and compared without relying on patients understanding the difference between a two or a three on a five-point scale of discomfort.
Mike Hollan is Pharm Exec’s Editor. He can be reached at firstname.lastname@example.org.