Do infectious diseases cause dementia? AI will be looking into it
UNSW Sydney
Working out the relationship between infectious diseases and dementia is a bit like the chicken and the egg problem. But artificial intelligence may help researchers decide which comes first.
For decades, the medical profession has viewed dementia as a progressive cognitive decline that typically affects older adults. While there have been advances in diagnosing it earlier and slowing its progression, there is no known cure.
But a new line of inquiry that first emerged in the 1990s is gaining ground among researchers, who are now exploring the possibility that infectious diseases might play a role in the development of this debilitating condition.
Dr Heidi Welberry with UNSW’s Centre for Big Data Research in Health and the Centre for Healthy Brain Ageing (CHeBA) will be examining these links not from a medical angle, but one informed by data science and biostatistics. She recently won an NHMRC Ideas Grant to use artificial intelligence to explore whether a causal link can be found between infectious diseases and dementia.
The use of AI in medicine is becoming increasingly common, such as in the design of new drugs, diagnosis of diseases, and targeted treatment tailored to a person’s genetic and physiological profile.
Recently, UNSW scientists successfully used AI to identify the chemical markers indicating early silicosis – a serious lung disease affecting construction workers who up until now, could only be diagnosed after irreversible damage turned up in CT scans. The same scientists had a couple of years earlier, developed an AI test that could predict, by analysing biomarkers in patients’ blood samples, which of them would go on to develop Parkinsons Disease.
In these cases the AI looks at correlations – it doesn’t matter so much whether the biomarkers are a cause of the disease or not; they just appear in patterns between affected individuals and assist with diagnosis. But when searching for causes, correlation isn’t enough.
“Mostly AI has been employed in tasks where the aim is prediction,” Dr Welberry says.
“Establishing causality is a different kind of task, because when you're trying to predict, you're looking for any signals that correlate with the disease - they don't have to be part of a causal pathway. The difference with our research is that we want to understand cause and effect. While someone may be hospitalised with a serious infection this t might be predictive of dementia, but it doesn't necessarily mean that it is causing the dementia.”
Infectious diseases and dementia
The theory that infectious diseases might be linked to dementia represents a shift in our understanding of its causes. Traditionally, dementia – of which there are about 100 types – has been attributed to damage in parts of the brain due to a range of possible causes including a combination of genetic factors, lifestyle factors, vascular damage and in the case of Alzheimer’s Disease the build-up of amyloid-beta plaques and other neuronal abnormalities. But emerging evidence suggests infectious diseases might also play a role through mechanisms like neuroinflammation and disruption of normal cell function.
“There’ve been quite a lot of studies that have shown a link between severe infections and subsequently developing dementia,” says Dr Welberry.
“There's also been quite a few studies that have shown protective effects of vaccinations. The study released this week showing the protective effect of the shingles vaccine on dementia provides the most important evidence to date. It made clever use of a health policy in Wales, UK that meant people who were born before 2 September 1933 were not eligible to receive the subsidised vaccine, whereas those who were born on or after this date were eligible. Comparing people born within a week or two either side of this date showed that vaccine coverage increased from 0.01% to 50% for those eligible and this corresponded to a 20% reduction in dementia incidence. There aren’t any other obvious explanations for this difference except that the vaccine is helping to prevent dementia.”
However, what we don’t know is how is the vaccine preventing dementia? Is it through prevention of shingles or is it a general activation of the immune system that helps the brain make repairs?
“We still don’t know for sure whether infections themselves lead to dementia,” Dr Welberry says.
“There are also other specific infections – including both viruses and bacterial infections – linked to an increased risk of dementia including pneumonia, syphilis, Lyme disease, gum disease, COVID-19 and HIV. It raises an interesting question: would preventing these diseases also protect against dementia?”
Chicken and egg
Epidemiologists like Dr Welberry examining the link between infectious diseases and dementia are faced with a chicken and egg problem: do infectious diseases create the conditions for dementia, or does dementia make the person living with the condition more susceptible to infection?
“In epidemiology, you would call that reverse causation, that the actual ordering of the events isn't always very easy to measure, particularly in observational data,” she says.
But one clue that supports the theory that infections may be causing the onset of dementia is in found in Alzheimer’s disease, the most common form of dementia.
“It might make sense that infections are part of the causal pathway, and particularly around the role of amyloid plaques – clumps of protein which appear in the brain of Alzheimer's disease patients. Recent research suggests the plaques might have a sort of antimicrobial effect as well, which means they could be part of the body’s defence mechanism against the infection, but unfortunately, the plaques also damage brain function.”
Where AI comes in
Dr Welberry says one of the ways she will use artificial intelligence to determine the plausibility of infectious diseases causing dementia is to compare a range of demographic, hospital, pharmaceutical and aged care data, between two groups of people, one that has been exposed to a particular infection, and one that hasn’t.
“We're going to try to mimic a randomised trial in the data,” she says. “So basically we’ll identify people who've been exposed to an infection and people who haven’t. And after balancing other factors that may account for the differences between the two groups, we’ll use the data to test the plausibility – using probability and statistics – that the infection caused dementia in the first group, versus other factors that could explain what’s happening.”
A matter of time
Making the task even more complex is the fact that dementia can be present long before it is medically diagnosed by a GP or a geriatrician. So what amount of time is acceptable between having an infection and developing dementia?
“Reverse causality is more plausible as an explanation if the infection and dementia diagnosis are quite close together in time. It may be that dementia was pre-existing and not diagnosed before the infection appears,” says Dr Welberry.
“So one of the things that we want to model is the testing of different potential lag times so that you can assess whether or not that is the plausible explanation.”
Without significant computer power, she says, crunching this much data and investigating all the relationships between the various variables collected over a long period would not be feasible.
Looking ahead, the team will be first trying to replicate the study on the Shingles Vaccine and Dementia here in Australia.
“We can conduct a similar experiment here with Australian data. The Shingles vaccine was introduced on 1 Nov 2016 in Australia with similar age restrictions to the Welsh study. We then aim to build on this by looking at casual paths in the data to better understand how the vaccine may be working to protect people. And we will also be looking at a lot of the other infections and what role they may play in the path to dementia.”
Contact details:
Lachlan Gilbert
UNSW News & Content
t: +61 404 192 367
e: [email protected]