Using artificial intelligence (AI) effectively may help to revolutionize the diagnosis, monitoring, and treatment of multiple sclerosis (MS), as well as optimizing understanding of the immune-mediated disease.
In the multiple sclerosis (MS) space, the use of artificial intelligence (AI) can translate into improvements across diagnoses, disease monitoring, and treatment selection, according to several experts from the Life Sciences Practice at global consulting firm Charles River Associates (CRA).
It involves effectively, and efficiently, using and focusing on 3 cognitive processes—learning, reasoning, and self-correction—to optimize outcomes.
AI is already hard at work in other disease states. For example, within HIV, it is being used to develop more accurate field-testing capabilities in rural South Africa. Developers from University College London and the Africa Health Research Institute applied deep learning algorithms to field-acquired HIV lateral flow test results for more accurate interpretation and classification as positive or negative. Results from their study show an overall 98.9% accuracy from using AI to classify test results compared with 92.1% seen when field workers used just their eyes.
Within kidney cancer, AI is being utilized via 2 prognostic risk-stratification models from Memorial Sloan Kettering Cancer Center and the International Metastatic RCC Database Consortium. The models are hoping to improve upon current methods of choosing optimal therapeutic combinations, particularly with the rise of immunotherapy-based combinations.
In addition, Bayesian Health recently published study findings on its Targeted, Real-Time Early Warning System (or TREWS) for sepsis, first deployed in 2019, which uses an algorithm within electronic medical records “to deliver accurate and actionable clinical signals that catch life-threatening events early, resulting in better patient health outcomes.”
However, a chief roadblock in correctly, and effectively, using AI for treating MS is thrown up by what some experts call “the bewildering number of symptoms it causes and the multiple ways in which they can present.” Some of these symptoms are fatigue, cognitive impairment, paralysis, bowel/bladder issues, and sexual dysfunction.
“Persistent challenges in the current approach for treating MS include that the underlying etiology of MS has yet to be characterized and a lack of reliable biomarkers,” stated Lev Gerlovin, vice president, Life Sciences Practice, CRA, in an interview with The American Journal of Managed Care® (AJMC®). “It is also difficult to make a timely and accurate diagnosis of MS because there are many possible causes of neurological symptoms. MS symptoms are also variable and unpredictable and can change over time, presenting challenges in monitoring patients and disease progression.”
Other conditions, too, are often confused for MS. These include Lyme disease, radiologically isolated syndrome, neuropathy, neuromyelitis optica spectrum disorder, fibromyalgia, Sjögren’s syndrome, myasthenia gravis, vitamin B12 deficiency, and acute disseminated encephalomyelitis.
“Symptoms of MS are unpredictable and vary in type and severity from one person to another, and symptoms can change in the same person over time,” stated Logan Wright, associate in the Life Sciences Practice at CRA, via an email exchange with AJMC®. “Common symptoms of MS include fatigue, numbness and tingling, blurred or double vision, weakness, poor coordination, imbalance, pain, depression, and problems with memory and concentration, which can mimic symptoms of other neurological disorders such as migraine, fibromyalgia, and neuromyelitis optica spectrum disorders.”
And according to Mayo Clinic, correctly diagnosing MS does not come from a test—because such a test does not exist specifically for that purpose. Instead, clinicians often utilize blood tests, a spinal tap, MRI, and evoked potential tests to differentially diagnose and rule out other conditions.
Throw in several common challenges associated with overall use of AI, and the picture becomes even more complex. These include the expensive computer power that the algorithms behind AI require, limited overall knowledge of how AI works, data privacy and security, potential bias, and scarcity of data.
However, none of these challenges have stopped AI from making inroads within MS.
AI is already being tested to gauge the potential of several biomarkers, such as levels of vitamins B12 and D3 and the mineral selenium, noted Andrew Thomson, consulting associate in the Life Sciences Practice at CRA, in an interview with AJMC®, “given that abnormalities in these vitamin levels are associated with a risk of developing MS. Or rather than focusing on biomarkers, AI may potentially allow clinicians to diagnose MS by classifying, quantifying, and identifying diagnostic patterns in medical images, including disease characteristics that may typically be too marginal for them to detect in a routine exam. Algorithms developed with AI technology can simplify tasks which are too labor intensive for clinicians and researchers.”
As a chemical messenger in the body, the National MS Society notes that studies have associated higher levels of vitamin D alone with a lower risk of developing MS attacks and MRI lesions, greater protection from developing the neurodegenerative disease, and a possibility to alter immune system function. In contrast, “low levels have also been associated with increased levels of disability.” The MS Society report additionally highlights the connection between Vitamin D and bone health, especially the “increasing awareness that low bone density (osteoporosis) may be underdiagnosed and undertreated in many people, including those with MS.”
C. Light Technologies is also using its eye-tracking technology to assess neurodegeneration via fixational eye motion mapping and predict MS disease progression. “With a device sensitivity down to 1 micron of movement, C. Light captures novel eye motion data to create a unique digital fingerprint of neurodegeneration,” the company wrote in a statement.
As recently as 2017, the yearly cost for medicines to treat MS averaged $70,000 without insurance coverage, according to NPR, despite the availability of generic drug formulations—so cost is always top of mind in the space. Even with appropriate coverage, patients still may need to pay thousands in out-of-pocket costs, noted Deborah Ewing-Wilson, DO, vice chair of operations, Neurology Department, University Hospitals Cleveland Medical Center.
When asked to predict how the use of AI may impact costs in the MS space, Thomson stated following initial start-up expenses, “the benefits will accrue over time. For example, a more efficient drug development process may help reduce costs for manufacturers and lead to more tailored treatment and better clinical outcomes, which may reduce treatment costs for both patients and health care systems. And if AI applications are delivered via telehealth versus an in-person visit, patient costs, including co-pays, may also be reduced.”
Having the potential to reverberate across the MS space, affecting not only drug developers and manufacturers, but patients, caregivers, and medical teams, too, AI needs to be both efficient and efficacious, Gerlovin, Wright, and Thomson concurred.
“AI has the potential to enhance the ways in which we diagnose, monitor, and treat diseases, including MS,” Gerlovin concluded. “Effective application of these technologies could lead to improved clinical outcomes at reduced costs, better patient care, and provide a platform for precision-based medicine, as well as expedite the development of the next generation of treatments.”