Guest Column | May 17, 2021

Harnessing The Power Of AI In MS Diagnosis, Monitoring, And Treatment

By Andrew Thomson, Logan Wright, and Lev Gerlovin, Life Sciences Practice, CRA

Brain Neuro

Applications of artificial intelligence (AI) are rapidly expanding in healthcare, including the handling of administrative logistics at hospitals, assessment of patient health data, and analysis of novel molecules in drug discovery. A prime example of the future potential of AI is in revolutionizing the way we diagnose, monitor, and treat multiple sclerosis (MS), thereby advancing our understanding of the disease course. Today, MS continues to be a difficult disease to diagnose and treat because the underlying etiology remains unknown. Persistent challenges in the current treatment approach to MS include a lack of reliable biomarkers, making patient diagnosis, disease monitoring, and drug discovery challenging. To maximize the utility of AI in MS, drug developers and medtech manufacturers must strategically incorporate AI solutions into their approach to MS drug development and commercialization.

Leveraging AI To Enhance The Diagnosis Of MS

Recent breakthroughs in AI research have demonstrated that computer-aided diagnosis can facilitate the early detection of MS via classification, quantification, and identification of diagnostic patterns in medical images. These enable earlier treatment interventions to reduce long-term MS-related disability. Data generated using AI techniques are analyzed automatically, taking the place of labor-intensive and time-consuming manual methods.

Diagnosis Disruption Factors: AI platforms may allow for diagnosis of MS through identification of indolent clinical characteristics that a physician may not otherwise notice. Earlier detection of MS has been shown to result in improved clinical outcomes and a reduced burden on the healthcare system.1 Developing these platforms requires the compilation of large clinical data sets to power the detection of MS patients. The ability to reference large reservoirs of clinical data can be used to power precision-based medicine, tailored to each patient, as MS subpopulations become evident.2 Accurate, cost-effective AI diagnostic tools for MS may allow physicians to treat more patients more effectively in the future, but it will be critical to understand where these streams of patient data are sourced and who manages them to effectively capture the required “building blocks” of AI from integrated health networks or large payer systems. These capabilities hinge on access to large patient databases, a potential barrier to development, as access to sensitive data remains a challenge for data privacy.

Diagnosis Case Studies

AI’s utility in specific applications shows how it can be employed to improve the diagnostic approach to MS:

  • Machine Learning Diagnostic Tools for MS Proof of Concept: S. Sharifmousavi and M. Borhani (2020) recently demonstrated that a support vector machine algorithm can accurately diagnose MS using patients’ plasma selenium, vitamin B12, and vitamin D3 counts.3 Aberrations in these vitamin levels are associated with a risk of developing MS and disease activity. The AI developed by these researchers was able to accurately differentiate between patients previously diagnosed with MS and control subjects, using only data on the blood levels of these biomarkers. This proof of concept establishes the potential of AI to elevate clinicians’ approach to diagnosing patients with MS.4
  • Eye Detection and AI Technology Potential to Identify MS: C Light Technologies, a neurotech and AI company, has developed an eye-tracking technology paired with machine learning that can be used to detect MS. C Light’s machine learning algorithms and instruments have potential to allow for earlier and more accurate prognosis of MS, leading to better patient outcomes and reduced overall healthcare costs.5

Better Monitoring Of MS Patients Using AI

Innovations in digital disease tracking technology may allow for the accumulation of real-world (RW) patient data in real time that can be integrated into burgeoning machine learning databases such as those being developed for MS. Advances in data collection technology, such as developments in biosensors and disease tracking applications for smartphones, unlock access to an unprecedented wealth of disease data. Using AI technology to sift through and derive meaningful patient care implications will be essential to capitalizing on these novel databases.

Disease Monitoring Disruption Factors: Physicians may use digital disease tracking technology to monitor their patients remotely, requiring in-office consultations only at precise milestones during disease progression and allowing for improved patient clinical outcomes and satisfaction. Another potential disruption factor of integrating AI into MS disease monitoring is the potential to broaden clinical understanding of the disease. In-office clinical consultations capture a snapshot of patients’ disease etiology, limiting the current treatment approach. Evolving toward the use of remote monitoring AI devices elevates the potential for superior clinical outcomes by capturing a patient’s comprehensive clinical profile, which may be leveraged for a more tailored treatment approach.

Monitoring Case Studies

Combined with advances in data tracking technology, specific developments in AI indicate future changes in MS disease monitoring:

  • Use of Biosensors and AI to Track Patient Disease Progression: Researchers from AbbVie and the University of California San Francisco used biosensors to track MS disease progression in Phase 2 clinical trials for patients receiving elezanumab, a monoclonal antibody currently under investigation to treat MS. The researchers believe that this RW data, gathered through patients wearing biosensors, could be leveraged to better capture patient prognosis, track their disease progression, and select appropriate clinical intervention.6
  • Remote AI Monitoring Device for MS: Outside of the academic setting, remote AI monitoring technology is beginning to gain traction among large drug manufacturers. Tools such as FLOODLIGHT MS, a smartphone-based digital assessment suite for MS in development by Roche and Genentech, are attracting attention for the way they can deliver novel measures to help detect if, and how, the underlying causes and symptoms of the disease are evolving.7

Improved Drug Discovery Platforms

Integration of AI approaches into pharmaceutical and biotech drug discovery platforms has the potential to expedite the identification of novel biologic therapies for challenging diseases like MS. Drug Discovery Disruption Factors: AI is set to disrupt the drug discovery process across multiple disease spaces, including MS, because of its ability to enable a rapid and less labor-intensive drug discovery platform.

Drug Discovery Case Studies

Specific advances in AI technology illustrate the potential for integration of this technology into current drug discovery platforms:

  • Use of AI with an In Vitro Model to Power MS Drug Discovery: AxoSim’s NerveSim, powered by Nerve-on-a-Chip, allows for the identification of neurological therapies for diseases like MS by leveraging an in vitro model to monitor the impact of drug candidates on the electrophysiological properties and cell-cell interaction of Schwann cells. This drug discovery platform more accurately represents human physiology, with the potential to reduce clinical failures and enable companies to develop effective drugs rapidly and at lower costs.8
  • AI Model to Identify Novel Biologic Drug Candidates: Another development has been the Alphafold 2 model from Google Deepmind, which recently demonstrated its ability to predict a protein’s 3D structure based on amino acid sequence at the Critical Assessment of Structure Prediction (CASP) challenge. During the CASP challenge, the Alphafold 2 model was found to be nearly as accurate as gold-standard experimental techniques, like cryo-electron microscopy, which are laborious and expensive.9

Conclusion

AI has the potential to revolutionize the ways in which we diagnose, monitor, and treat MS. Effective application of these technologies could lead to improved clinical outcomes at reduced costs, provide a platform for precision-based medicine, and expedite the development of the next generation of MS treatments. Drug developers and medtech companies must understand how to incorporate AI into their drug discovery platforms and approaches to developing diagnostic tools, especially in nuanced disease areas like MS. To do so, the sources of required patient and market data must be accessible to all stakeholders as the building blocks of AI. If these data can be consolidated and accessed across the value chain, AI technology can become the focal point of the drug discovery process and the treatment paradigm for MS. While it is clear that AI will be a critical factor in drug development and treatment of MS, the integration of AI into healthcare is not limited to MS. Drug developers operating in other diseases, particularly those with high unmet needs, poorly characterized underlying pathways, and expensive and burdensome drug discovery approaches, must be able to understand and harness the utility of AI to be successful in the coming years.

References

  1. Waubant, Emmanuelle. “Improving outcomes in multiple sclerosis through early diagnosis and effective management.” The primary care companion for CNS disorders vol. 14,5 (2012): PCC.11016co2cc. doi:10.4088/PCC.11016co2cc.
  2. Brasil, Sandra et al. “Artificial Intelligence (AI) in Rare Diseases: Is the Future Brighter?” Genes vol. 10,12 978. 27 Nov. 2019, doi:10.3390/genes10120978.
  3. Sharifmousavi, Seyed Sajjad, and Matia Sadat Borhani. “Support Vectors Machine-Based Model for Diagnosis of Multiple Sclerosis Using the Plasma Levels of Selenium, Vitamin B12, and Vitamin D3.” Informatics in Medicine Unlocked, vol. 20, 2020, p. 100382., doi:10.1016/j.imu.2020.100382.
  4. Sintzel, Martina B et al. “Vitamin D and Multiple Sclerosis: A Comprehensive Review.” Neurology and therapy vol. 7,1 (2018): 59-85. doi:10.1007/s40120-017-0086-4
  5. C. Light Technologies, www.clighttechnologies.com/.
  6. Pfleeger et al. 2020. “Novel Digital Outcomes in Multiple Sclerosis Clinical Trials: Use of Wearable Biosensors to Collect Real-World Subject Activity Data in Two Phase 2 Studies of Elezanumab in Multiple Sclerosis.” Presented at the 8th ACTRIMIS-ECTRIMS MS Virtual 2020 Meeting.
  7. FloodlightOpen, floodlightopen.com/en-US.
  8. AxoSim, 15 Oct. 2020, axosim.com/nervesim/.
  9. Callaway, Ewen. “'It Will Change Everything': DeepMind's AI Makes Gigantic Leap in Solving Protein Structures.” Nature News, Nature Publishing Group, 30 Nov. 2020, www.nature.com/articles/d41586-020-03348-4.

About The Authors:

AuthorAndrew Thomson is a consulting associate within the Life Sciences Practice at CRA with more than four years of experience in commercial strategy consulting with pharmaceutical and biotech clients. His background stems from the strategic and technical sides of the space, with expertise in strategic development and quantitative analysis, and focused on commercializing disruptive healthcare products, including AI-based surgical robotics, microbiome-based diagnostics, and gene therapies.

 

AuthorLogan Wright is an associate within the Life Sciences Practice at CRA with more than three years of experience in pricing and market access strategy consulting with firms across the life science industry. His work has focused on assisting drug manufactures with their go-to market pricing and access strategy and development of market opportunity assessments for pipeline assets across therapy areas, including the rare disease space.

 

AuthorLev Gerlovin is a vice president in the Life Sciences Practice at CRA with more than 14 years’ experience in life sciences strategy consulting focused on commercial and market access strategies. He has advised biopharmaceutical and medical device manufacturers on go-to market as well as lifecycle management strategies, and has supported several assets focused specifically on patient care in multiple sclerosis. Lev also heads up CRA’s Life Sciences 2030 Idea Leadership Platform.

The views expressed herein are the authors’ and not those of Charles River Associates (CRA) or any of the organizations with which the authors are affiliated