Amidst the complexities of neurodegenerative diseases, artificial intelligence is emerging as a possible game changer in designing experiments and hunting for a cure.
Artificial intelligence (AI) is at the beginning of a revolution and could create the biggest change in how the world works since the emergence of the internet. The question is, how can it help us untangle the mysteries behind some of these seemingly incurable diseases?
A review paper led by Dr. Marzi at Imperial College London delves into the potential of using AI in creating experimental models for neurodegenerative diseases. The review looks to demonstrate the current challenges and possible solutions of AI- and Machine Learning (ML)-based approaches that could be applied to enhance research.
The paper highlights various genetic variant prediction models like Enformer, Basenji, and a semi-supervised deep learning approach (CNN-GNN), that show improved performance when trained on data from multiple species rather than a single species. The study emphasises the importance of leveraging diverse species data to enhance research outcomes in neurodegenerative diseases like Alzheimer’s and dementias.
Not so fast
Reproducibility is a major issue both within individual model systems and across species used in research. One such model, human-induced pluripotent stem cells (iPSCs), is now a critical component in the researcher’s toolbox for evaluating human disease. It allows for more accurate modeling than traditional immortalised cell lines, while still allowing for high throughput investigations compared with in vivo models. However, the Human Induced Pluripotent Stem Cells Initiative (HipSci) reported that 5–46% of phenotype variation is due to individual genetic background, which can lead to huge variability in results.
Laboratory and protocol-based sources of variation can overpower genotypic effects and lead to further issues with reproducibility. Similarly, while no in vivo mouse model can fully encapsulate all aspects of a disease, even isolated mechanistic models can have issues with reproducibility.
Why does it matter
Dementia is a leading cause of disability and dependency among older adults worldwide. According to the World Health Organization (WHO), in 2019 the global statistic for people with dementia was around 55 million and this number is expected to increase annually by around 10 million. Additionally, the global economic burden of dementia for health and social care expenditure was 1.3 trillion US dollars.
Incorporating AI and ML models into dementia research could offer numerous benefits and advancements in understanding, diagnosing and treating the condition. One of the challenges in dementia research is identifying the condition at an early stage and when interventions might be most effective. ML algorithms can be trained to recognise subtle patterns in the research models that could indicate early-stage dementia, enabling earlier diagnosis and intervention. It is hoped that in the future, AI may be able to successfully automate repetitive tasks for more efficient use of research time.
Take home messages
1. There has been significant progress in the application of AI to experimental models of neurodegenerative diseases, such as Alzheimer’s disease.
2. Despite the potential of AI in research, many reproducibility issues exist that need to be resolved within models and across species.
3. If AI and ML models can learn to identify dementia at an early stage, the technology could enable faster diagnosis and treatment for patients.
Guest author: Conor McQuaid, PhD
This article was written as part of a series of ‘journal club’ summaries for Scientific Writers Ltd., and is based on the following publication:
Title: Artificial intelligence for neurodegenerative experimental models.
First Author: Sarah J. Marzi et al.
Journal: Alzheimer’s & Dementia
Date online: 28th September 2023
Other references:
https://www.who.int/news-room/fact-sheets/detail/dementia
Thanks for reading this review summary. We hope you found it informative and if you’d like to be reminded when each summary is published, hit the ‘subscribe’ button. We appreciate your support.
