Comprehensive Summary
This article by Hillis et al. explores the impact of artificial intelligence (AI) on the prognostic features of neurology. AI has been increasingly used to assist in the diagnostic procedures, detecting neurological diseases through machine learning models and neuroimaging models. However, this article suggests that although AI is very well engrained in the diagnostic processes of neurology, its application in prognosis is still evolving. In early 2024, the Food & Drug Administration (FDA) authorized the first AI/machine learning (ML) device to monitor the prognosis of mild cognitive impairment (MCI) to see if it could progress to dementia or Alzheimer's disease.(AD) This device was recently granted a de novo request, marking a significant milestone of AI in neurology prognosis. However, the FDA has provided multiple risks with the device, including incorrect predictions, which could lead to significant health concerns. To mitigate these risks, the FDA is implementing a special control using post-market surveillance to test the product in a real-world setting in a mixture of demographics. Overall, this device, although in its infancy, has the potential to assist in the prognosis of neurological diseases, with it being able to integrate MRI scans, demographic information, and medical assessments to create a detailed prognosis. Although there are still many challenges, the rapid evolution of AI in prognosis proves to be a promising advancement.
Outcomes and Implications
The authorization of this AI/ML-based prognostic device will offer a new tool for clinicians to assess the likelihood of an MCI progressing into AD or dementia. While this device’s ability in neurological prognosis is promising, clinicians still need to be careful in interpreting these predictions, since the accuracy of this device is still under clinical evaluation. However, the device’s multimodal approach proves promising for enhancing the clinical capabilities, allowing for more of the prognosis to be attributable to a specific cause. The FDA’s post-market surveillance is important to prove the device’s accuracy in a wide variety of real-world scenarios, however this device has the potential to be used in a multitude of neurological disorders to predict prognosis and causes of these deficits.