Blog – Enhancing Dementia Drug Discovery with AI

Artificial Intelligence (AI) is becoming an ever growing presence across all aspects of society. It seems like pretty much every industry is calling out for AI based solutions to help streamline critical processes. Dementia research is no different and we had an excellent blog earlier this month from Ajantha Abbey about using AI to enhance dementia diagnosis. I highly recommend you read or listen to it. Diagnosis is one of the biggest challenges in dementia care and any technology that can potentially improve this process is not just of interest. It is critical to progress.

But what about the other major aspect of dementia research? Can we harness the power of AI to improve our ability to discover and develop new therapeutic targets for dementia? The short answer is yes. The longer answer is yes… If we use it correctly (and by that we mean ethically as well as scientifically).

Automating Data Analysis

The value of AI in dementia research mostly lies in the capacity to use it to analyse huge quantities of data and identify patterns or correlations. Time is of the essence in dementia research. Across the globe, someone develops dementia every 3 seconds. It’s hard not to feel that clock constantly ticking away when you are sat spending hour upon hour trying to analyse a gigantic genomic or proteomic data set. It is in situations like this where we can enlist the help of AI to get us results much quicker.

In particular, we can harness the power of machine learning techniques to develop predictive algorithms that can screen vast quantities of data and identify patterns and correlations. Essentially, they can do the type of analysis you would do yourself but in a fraction of the time (and probably with greater accuracy).

For example, say you have a large volume of proteomic data from the brains of post-mortem dementia patients. You want to analyse this data for any correlations between specific proteins and the manifestation of dementia. A correctly trained machine learning algorithm can run that analysis for you, identify the proteins that are strongly associated with dementia and even screen them against the ones we already know about so that you only identify potentially new therapeutic targets.

This isn’t just an idea too. It is a reality that is playing out right now at institutions like the University of Cambridge. It’s really exciting stuff! https://www.cam.ac.uk/research/news/ai-driven-techniques-reveal-new-targets-for-drug-discovery

Drug Repurposing

Another exciting thing about employing AI in dementia research is that it is so adaptable. We have talked about automating data analysis and that is often at the core of how it is used. However, the data you choose to analyse can be literally anything. It can be medical imaging, protein expression, genomics or even databases on drug action. During the pandemic we heard a lot about ‘drug repurposing’ – finding another application for a drug that is currently in clinical use. The advantage of this approach is it can speed up the clinical approval of a therapy as a pre-approved drug for another indication can fast track to phase 2 clinical trials for the new therapy area.

There could be a multitude of drugs in the clinic right now that could provide some form of relief or benefit to dementia patients. Even the smallest improvement would be welcomed by anyone fighting against dementia. The difficult part is identifying which drugs could be of use as it requires a lot of understanding behind how each individual medicine works and how that could translate to dementia.

This is again where AI can provide us with huge gains. We can use it to expedite that process of identifying candidates for repurposing by analysing the biological pathways of the drug and evaluating if it aligns with any biological mechanisms in dementia. There’s actually a really nice proof of concept study from a group at the Vanderbilt University Medical Center that came out in Nature earlier this year. https://www.nature.com/articles/s41746-024-01038-3

Personalising Treatment

Diving deeper into the angle of drug discovery, we can also use AI to better personalise the use of therapies for dementia. As a species we are incredibly diverse and that diversity also manifests in the mechanisms that drive disease. Two patients with the same dementia causing illness can have completely different genetic, metabolic, and environmental driving factors. Expecting to give them exactly the same therapy to generate exactly the same response seems unrealistic.

In an ideal scenario we would give both patients treatment that is fully personalised based on factors such as their genomic profile and biomarker expression. In reality, no healthcare system in the world is currently equipped to deal with the amount of processes required to deal with this. But what if they had help from AI-driven predictive modelling to rapidly analyse these diverse datasets? Suddenly that burden is hugely reduced. Personalised medicine has always been this holy grail that we have marched out in search of and AI could potentially help us get there!

Aiding Clinical Trials

Finally, can we use AI to improve the single biggest blockade against the clinical translation of new therapies – clinical trials? Clinical trials are eye-wateringly expensive and eliminate 90% of all promising new drug candidates. That means 90% of the money that is spent on clinical trials results in failure. We need to find ways to do better. The integration of AI in the trial process could offer promising solutions to overcome the giant obstacle of clinical translation. Drugs fail trials for a multitude of reasons. The best way to avoid failure is to identify the reason your therapy might fail BEFORE it does so you can address it.

That is where AI comes in. If we can use it for all stages of the trial including selection of suitable patients, prediction of outcomes and virtual trials to help us design the ‘optimal’ clinical study parameters, we may be able to improve that success rate and get more drugs to the patients that need them.

Keeping it Ethical

But is any of this ethical? This is the biggest question that underlies hesitance around adopting AI. People want to know that the dangers are considered and we are being cautious where needed. We must carefully consider the ethical implications of using AI in drug discovery, especially when using it will involve handling sensitive health data. Maintaining patient confidentiality is a must and all studies using patient data must take every step to ensure they have informed consent before they go ahead. I believe AI has the potential to massively improve drug discovery in dementia research so lets make sure we do everything properly so we can actually use it to its full potential.

Dr Sam Moxon

Author

Dr Sam Moxon is a Research Fellow at the University of Birmingham. His expertise falls on the interface between biology and engineering. His PhD focussed on regenerative medicine and he now works on trying to develop 3D bioprinting techniques with human stem cells, so that we better understand and treat degenerative diseases. Outside of the lab he hikes through the Lake District and is an expert on all things Disney.

Follow @DrSamMoxon