Neurology/Technology: An Essential Intersection

By: Kate Constan, Cognitive Science ‘ 2026

Ironically, computers seem to have become the focal point of life science discussions: machine learning and artificial intelligence (AI) are hot topics within the world of neuroscience. Even just flipping through the Healthcare Review, readers will find article after article about AI’s medical capabilities. When asking how neuroscientists and neurologists can best use machine learning, the answer may lie in AI's diagnostic abilities. 

There are exciting and emerging illustrations of the power of AI to improve medical care. In early January 2024, an article in Nature Communications documented that severe multiple sclerosis (MS) can be predicted by machine learning when given a combination of proteins found in cerebrospinal fluid. The authors of “Proteomics reveal biomarkers for diagnosis, disease activity and long-term disability outcomes in multiple sclerosis” use a protein-forward approach to MS diagnosis [1]. In this study, Åkesson et al. quantified 1463 proteins in the cerebrospinal fluid of MS patients and healthy controls and longitudinally followed these individuals. By feeding the biomarkers and patient outcomes into a computer program capable of learning from patterns and creating algorithms, Åkesson et al. discovered that this machine-learning system could accurately predict disease progression [1]. It is difficult for doctors to give an MS patient an accurate and timely prognosis, so the boost from AI is substantial. With AI's help, doctors can provide better information, and MS patients can plan better for the future. Between setting up accessible solutions at home, arranging long-term healthcare, and choosing the correct medication, better planning fosters better quality of life.

Earlier papers about Alzheimer's Disease showed similar prognostic abilities. In 2022, a consortium of brain scientists used machine learning to sort typical brain aging from disease-driven cognitive decline. “Disentangling Alzheimer’s disease neurodegeneration from typical brain aging using machine learning” discusses strategies for adequately diagnosing Alzheimer’s Disease (AD), which is sometimes difficult to distinguish from non-pathological cognitive decline [2]. Hwang et al. assessed 4054 MRI scans from AD patients and controls, looking for neuroanatomical biomarkers. Similarly to Åkesson et al., these authors effectively taught a machine-learning model to discern AD from typical aging [2]. While this is not accurate every time, physicians can incorporate it into their input to help inform a treatment plan [3].

Doctors and AI are not mutually exclusive. Instead, physicians and machines can aid in patient support in a cycle of diagnostic information. An article published by Harvard Medical School suggests that “AI will likely empower the practice of medicine” rather than kill the  doctor-patient relationship [3]. Empathy is essential to treatment; patients would rather hear about a diagnosis from a trusted doctor than from some new, mysterious digital source [3]. M

Machine learning and data analytics can be integral for precise and effective treatment, but they cannot replace human compassion, and often require human assistance. Despite causes for optimism about the ability of AI to help diagnose neurodegenerative disorders and beyond, the medical industry must exercise caution. An informed approach to incorporating AI is essential. In Alzheimer’s disease detection, machine learning models are not yet precise enough to ensure accuracy. While doctors make errors, and patients often seek second opinions, AI alone is not the answer. There is work to be done with improving AI diagnostics [4]. In a review of machine-learning-based neurological diagnosis, Dara et al. suggest adding additional variables to datasets to give AI a fuller depiction of patient health before allowing it to draw conclusions [5]. MRI, PET, and genetic data can all be combined to help mitigate overgeneralization issues in AI. By optimizing and refining the use of AI in Alzheimer’s diagnosis, we can mitigate errors.

In the neurodegenerative sphere and beyond, the patient focus is essential. Early and effective diagnosis of Alzheimer’s, Parkinson’s, and ALS can help families plan for the future and provide the best care. Doctors are essential for maintaining a high level of care, which should include responsibly interpreting AI-generated information. While deep learning models can analyze images and biomarkers to diagnose better than human practitioners, patients require competent treatment from every angle. Essential steps include creating more interpretable models, finding larger sample sizes, and utilizing additional data points. Medicine’s new frontier is bright, but certain issues loom. Is trusting human care to AI ethical? Who will be held liable for AI errors? If patient outcomes are maintained as a priority, conversations surrounding these issues cannot be rushed. In neurology and beyond, AI has driven new opportunities as well as new concerns. As AI is in its nascence, this is only the start. Is neurological diagnosis ready for a technological upheaval? 

References

1. Julia Åkesson, Sara Hojjati, Sandra Hellberg, Johanna Raffetseder, Mohsen Khademi, Robert Rynkowski, Ingrid Kockum, Claudio Altafini, Zelmina Lubovac-Pilav, Johan Mellergård, Maria C. Jenmalm, Fredrik Piehl, Tomas Olsson, Jan Ernerudh, Mika Gustafsson. Proteomics reveal biomarkers for diagnosis, disease activity and long-term disability outcomes in multiple sclerosis. Nature Communications, 2023; 14 (1) DOI: 10.1038/s41467-023-42682-9

2. Gyujoon Hwang, Ahmed Abdulkadir, Guray Erus, Mohamad Habes, Raymond Pomponio, Haochang Shou, Jimit Doshi, Elizabeth Mamourian, Tanweer Rashid, Murat Bilgel, Yong Fan, Aristeidis Sotiras, Dhivya Srinivasan, John C. Morris, Marilyn S. Albert, Nick R. Bryan, Susan M. Resnick, Ilya M. Nasrallah, Christos Davatzikos, David A. Wolk, from the iSTAGING consortium, for the ADNI, Disentangling Alzheimer’s disease neurodegeneration from typical brain ageing using machine learning, Brain Communications, Volume 4, Issue 3, 2022, fcac117, https://doi.org/10.1093/braincomms/fcac117

3. James, T. A. (2023, April 13). How artificial intelligence is disrupting medicine and what it means for physicians. Harvard Medical School Postgraduate Education. https://postgraduateeducation.hms.harvard.edu/trends-medicine/how-artificial-intelligence-disrupting-medicine-what-means-physicians

4. Arya, A. D., Verma, S. S., Chakarabarti, P., Chakrabarti, T., Elngar, A. A., Kamali, A. M., & Nami, M. (2023). A systematic review on machine learning and deep learning techniques in the effective diagnosis of Alzheimer's disease. Brain informatics, 10(1), 17. https://doi.org/10.1186/s40708-023-00195-7

5. Dara, O. A., Lopez-Guede, J. M., Raheem, H. I., Rahebi, J., Zulueta, E., & Fernandez-Gamiz, U. (2023). Alzheimer’s disease diagnosis using machine learning: A survey. Applied Sciences, 13(14), 8298. https://doi.org/10.3390/app13148298