Dementia and AI

 Today, more than 55 million people worldwide have been diagnosed with dementia and it is believed that the number will increase by 10 million every year according to new data published by the World Health Organization (WHO)1. Dementia has also become the 7th leading cause of death worldwide2. This make it one of the biggest healthcare challenges we face as a society. A healthcare challenge which is still surrounded by a great deal of stigma.

Every September is World Alzheimer’s Month, in order to destigmatize dementia and raise awareness around Alzheimer’s Disease (AD) and dementia3. Organizations all over the world prepare educational resources and other activities to provide information and support to patients and families4.

With that in mind, we wanted to support the cause creating this guide to show the latest developments of artificial intelligence (AI) in the field of Dementia and how AI can support the whole diagnostic process as well as the day-to-day life of people living with dementia or AD.

Dementia and AD

Before we dive into the different applications of AI for dementia, let’s summarize dementia.

Dementia is characterized by atrophy of different regions of the brain, and with that, deterioration of cognitive function greater than the one expected from normal ageing. Depending on the specific regions that are most affected, we can encounter different types of dementia, AD being the most common form as it represents 60% to 70% of the diagnosed cases5.

Can we treat dementia?

Although there has been research and development regarding medication to mitigate dementia6, currently there is no cure. Because of that, the treatment route chosen by specialists all over the world are strategies to help mitigate the symptoms, improve the quality of life, and offer support and care to those affected by dementia5.

How does dementia impact the life of those affected and their surroundings?

Dementia not only deeply affects the lives of those who suffer from it, but it also heavily impacts the lives of their support system. The amount of care needed can become very disruptive for families and friends as they are often the ones that become principal caregivers. They can spend, on average and at least in the first stages,  up to 5 hours per day attending to their loved ones5.

As a result, the process can be very overwhelming and stressful, physically, mentally, and financially5 for both patient and family and can lead in some cases to less-than-ideal environments for the necessary care.

AI applications to dementia

For these reasons we believe in the importance of education and research to improve the process and the quality of lives of those affected by dementia. When it comes to dementia, and particularly, AD, there is ample research that applies AI algorithms to the clinical pathway with different goals. For example, improving early diagnosis through automating, standardizing and improving the accuracy of the predictions, or developing effective treatments7.

AI for a better diagnosis of dementia 

As we have covered above, there are different types of dementia that have different symptoms, prognosis and treatment. For that reason, an accurate and early diagnosis will improve the quality of care of those who suffer. When it comes to the diagnostic process, there has been plenty of research evaluating the use of AI models and algorithms for different diagnostic paths.

AI applications to AD were first explored in the field of neuroimaging. Throughout the years, researchers have focused on trying to extract as much information as possible from brain scans in an effort to track volume changes in the brain, and therefore, track brain atrophy8,9. Some early examples include an AI algorithm that was able to classify AD based on MRI scans with the accuracy of 92.36%10 or an AI algorithm that was able to give a prediction more than 75 months earlier than the final diagnosis with 82% specificity and 100% sensitivity11.

Besides neuroimaging, neurologists use certain cognitive tests during consultations and yearly checkups to assess symptoms or conditions, such as aphasia, related to dementia and AD12. AI research has been conducted with the goal of making cognitive tests13, speech assessments14, and dementia screenings like the clock drawing tests15 reproductible on a larger scale, so that they can reach a larger, even remote, populations.

Clinical implementation

Today, the development of these type of algorithms has reached clinical practice with multiple available AI solutions cleared by regulatory bodies like the FDA.

Discover Quantib's FDA cleared and CE-marked AI-driven software for brain atrophy and WMH quantification. 

QUANTIB ND

However, researchers16 believe there are 3 areas that need some work in order to implement these AI models: the questions defined for the algorithms to solve need to be clear and relevant to the memory clinic; the heterogeneity of the algorithms make it very difficult to compare their performance, and the lack of available of representative data to test the algorithms.

AI for a better prognosis of dementia 

Prognosis has been a large part of AI research into AD. It primarily looks at the progression of mild cognitive impairment to AD7 with two different approaches, either using single-time-point data or doing a longitudinal analysis of imaging scans16. The results differ greatly, however, one specific study was able to conclude that using longitudinal data provide better accuracy (88.61%) than using fixed baseline data (78.48%)17.

AI software for treatment research

As mentioned before, there is no cure for dementia or AD. At the moment. However, it is believed that there are different ways in which AI could promote the research involved in potential pharmacological treatments, that generally aim to improve symptoms and delay progression of the disease.

AI has the potential of creating tailored treatments for different clinical phenotypes of dementia and AD patients through predictive models that include expertise from an array of different fields, from neurology and clinical signs, to geriatric medicine and statistics7. This might allow prediction of which medication works best for each individual patient. Researchers also believe AI software has a lot of potential to select drugs to be used for the treatment of AD, for example by attacking the effects the disease has in the brain (e.g., plaques)7,18,19.

One reason for the limited amount of treatment options might be the difficulty within clinical trials to enroll early dementia sufferers. Most participants in clinical trials are too far along in the disease for differences to be easily perceived9. AI could help clinical trials not only by providing earlier diagnosis and therefore enrolling patients that might yield better outcomes, but also by simplifying the selection and/or the implementation of those clinical trials thanks to optimizing steps such us participant selection or real-time diagnostics7.

AI software for post-diagnostic support

It is believed that AI can also be used after the AD or dementia diagnosis, for example, by creating a smart environment that can monitor the person’s behavior and movement 13 and can alert healthcare professionals to help when it is needed20.

The healthy home adapted from UK Dementia Research Institute, 202220.

This is said to improve the quality of life of those affected by dementia and their families as it can potentially ensure a safe environment in which they can stay independent and in their own homes for longer.

AI and technology to support day-to-day tasks

The use of virtual home assistants, Like Google Home or Google Assistant, or even smartphones allow the personalization of reminders and alarms that can be extremely useful to avoid people forgetting important daily tasks, such as going to the toilet, having a drink or taking their medication21.

Conclusion

The already dramatic number of cases of dementia is expected to increase by millions each year. Due to the characteristics of the disease and its burden, it is of the utmost importance to find new ways in which we can improve each step within the clinical pathway of dementia. This could allow them to live with dignity, and greatly improve the day-to-day life of patients and their loved ones.

We believe that further research utilizing AI can advance the field and ultimately provide better support to those who need it.

Bibliography

1. World Health Organization (WHO). Global Dementia Observatory (GDO). https://www.who.int/data/gho/data/themes/global-dementia-observatory-gdo (2021).
2. World Health Organization (WHO). The top 10 causes of death. https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death (2020).
3. Alzheimer’s Disease International (ADI). Dementia statistics. https://www.alzint.org/about/dementia-facts-figures/dementia-statistics/ (2020).
4. Alzheimer’s Disease International (ADI). World Alzheimer’s Month. https://www.alzint.org/get-involved/world-alzheimers-month/ (2022).
5. World Health Organization (WHO). Dementia. https://www.who.int/news-room/fact-sheets/detail/dementia (2021).
6. Arthur Post Uiterweer. How Aducanumab will impact neurology and neuroradiology practice. Quantib How Aducanumab will impact neurology and neuroradiology practice (2021).
7. Fabrizio, C., Termine, A., Caltagirone, C. & Sancesario, G. Artificial Intelligence for Alzheimer’s Disease: Promise or Challenge? Diagnostics 11, 1473 (2021).
8. Qiu, S. et al. Multimodal deep learning for Alzheimer’s disease dementia assessment. Nat Commun 13, 3404 (2022).
9. Brierley, C. AI could detect dementia years before symptoms appear. University of Cambridge (2021).
10. Zhang, Y. et al. Detection of subjects and brain regions related to Alzheimer’s disease using 3D MRI scans based on eigenbrain and machine learning. Front Comput Neurosci 9, (2015).
11. Ding, Y. et al. A Deep Learning Model to Predict a Diagnosis of Alzheimer Disease by Using 18 F-FDG PET of the Brain. Radiology 290, 456–464 (2019).
12. Dashwood, M., Churchhouse, G., Young, M. & Kuruvilla, T. Artificial intelligence as an aid to diagnosing dementia: an overview. Prog Neurol Psychiatry 25, 42–47 (2021).
13. Li, R. et al. Applications of artificial intelligence to aid early detection of dementia: A scoping review on current capabilities and future directions. J Biomed Inform 127, 104030 (2022).
14. Fristed, E. et al. Evaluation of a speech-based AI system for early detection of Alzheimer’s disease remotely via smartphones. doi:10.1101/2021.10.19.21264878.
15. Amini, S. et al. An Artificial Intelligence-Assisted Method for Dementia Detection Using Images from the Clock Drawing Test. Journal of Alzheimer’s Disease 83, 581–589 (2021).
16. Borchert, R. et al. Artificial intelligence for diagnosis and prognosis in neuroimaging for dementia; a systematic review. medRxiv 2021.12.12.21267677 (2021) doi:10.1101/2021.12.12.21267677.
17. Teng, L. et al. Predicting MCI progression with FDG-PET and cognitive scores: a longitudinal study. BMC Neurol 20, 148 (2020).
18. Schneider, G. Automating drug discovery. Nat Rev Drug Discov 17, 97–113 (2018).
19. Mishra, R. & Li, B. The Application of Artificial Intelligence in the Genetic Study of Alzheimer’s Disease. Aging Dis 11, 1567 (2020).
20. UK Dementia Research Institute. Launching: UK DRI Care Research & Technology at Imperial. https://ukdri.ac.uk/news-and-events/uk-dri-care-research-technology-at-imperial (2022).
21. Topsfield, J. Robot dogs and Google reminders: How AI helps people with dementia stay independent. The Sydney Morning Herald (2022).