Technologies and Therapies: The Intersection of Innovative Diagnostics and Alzheimer's Treatment

In 1906, Dr. Alois Alzheimer, a German physician, provided the first description of what he called "a peculiar disease" characterised by profound memory loss and microscopic brain changes. This condition, now known as Alzheimer's disease (AD), accounts for 60% to 80%, of all dementia cases.¹ Alzheimer's is a degenerative disease, with symptoms manifesting gradually and intensifying over time.

Globally, the incidence of AD and other dementias saw an astounding nearly 150% increase between 1990 and 2019,² reflecting the growing burden it places on individuals, families, and healthcare systems. Currently affecting 6.1 million people in the United States alone, projections paint a grim picture with an expected rise to 13.8 million cases by 2060.³ Additionally, the number of AD patients worldwide is expected to exceed 150 million by 2050.⁴ This underscores the urgent need for continued research, awareness, and support to alleviate the profound burden this disease places on society.

The application of artificial intelligence (AI) in Alzheimer's care represents a groundbreaking shift in managing this challenging disease. For patients, AI can mean more accurate diagnoses and personalised care strategies, while for caregivers, it introduces efficiency in monitoring patient health and behaviour, thus potentially reducing the burden of care. The integration of AI into Alzheimer's care is not only enhancing the quality of life for those affected but also transforming the landscape of medical research and healthcare delivery in neurodegenerative diseases.

Improving quality of life

While the pursuit of treatments to slow disease progression and ultimately find a cure for dementia is undeniably crucial, it's equally imperative to prioritise the enhancement of the quality of life (QoL) for individuals living with this condition. Often, as the disease advances, people with Alzheimer's disease or other types of dementia find themselves in retirement homes and psychiatric facilities, where maintaining their autonomy and sense of identity becomes increasingly challenging. The loss of privacy and the inability to sustain their familiar lifestyle and engage in leisure activities can significantly diminish their QoL.⁵ Additionally, mobility issues, which can result from dementia itself or coexisting health conditions, further compound the challenges faced by individuals with dementia.⁶ As a result, many are constrained from participating in activities and experiences they once relished, highlighting the urgent need for comprehensive care and support strategies that prioritise their well-being and preserve their dignity.

An innovative approach to enhancing the quality of life for individuals with dementia involves the use of virtual reality (VR). VR immerses users in computer-generated, interactive, three-dimensional environments through a head-mounted display, creating a sensation of actually being in the virtual world. In a 2019 study conducted in a locked psychiatric facility with dementia patients, participants were given the choice of various VR locations, including a cathedral, forest, rocky beach, sandy beach, and countryside.

The findings revealed that VR treatment holds significant potential to improve the QoL for people with dementia.⁷ During and for a short time after their VR experiences, participants exhibited improved moods, with one participant expressing, "It was the best day," and sharing their enthusiasm with peers. Notably, participants could recollect old memories, engage in activities that were otherwise challenging, and even fostered better relationships with caregivers, prompting discussions about their lives and memories. This breakthrough not only enhances the caregiver-patient relationship but also reduces aggression and offers a sense of personal space, underlining the potential of VR as a valuable tool in dementia care.

In the future, the use of VR in dementia care could be expanded to include more personalised experiences within this virtual world. Instead of using common locations such as a beach or a park, VR could be used to explore a patient's old home or a place they enjoyed visiting in the past, aiding in unlocking some happy memories, even if only temporarily.

Detection and diagnosis

The pre-symptomatic stage of Alzheimer's disease is the key window for early detection and successful intervention.⁸ By improving the detection and diagnosis of AD, preventative measures can be implemented earlier, reducing the burden of this disease.

One promising non-invasive AD biomarker is blood-based data. This data is typically generated using omic techniques (genomics, proteonomics, transcriptomics) which produce vast and complex datasets with high dimensionality that make manual analysis impractical and inefficient. For instance, a typical transcriptomics dataset may contain readings from thousands to tens of thousands of samples, such as cells or tissues, including data on up to around 20,000 genes, creating a labyrinth of information.⁹

To navigate this complexity, AI, which excels in processing large and intricate datasets, has been employed effectively. Researchers have crafted a tree-based machine learning algorithm specifically tailored to analyse transcripts in blood.⁹ This innovation has led to the discovery of small sets of transcripts that can be used to differentiate between healthy individuals and those with AD or other neurodegenerative diseases. The algorithm achieves this with a high degree of sensitivity and specificity, marking a significant stride toward early and accurate diagnosis of AD through non-invasive means.

Additionally, recent study has made a significant breakthrough in the early detection and progression tracking of Alzheimer's disease (AD) using advanced machine learning techniques applied to neuroimaging.¹⁰ Researchers developed a deep learning (DL) model trained on MRI scans from both healthy individuals and AD patients to create a "deep learning MRI score" that quantifies the probability of AD in each scan. Researchers identified a cohort of individuals with mild cognitive impairment (MCI) at the outset, who had complete sets of cerebrospinal fluid (CSF) amyloid and tau biomarkers and structural MRI scans.

The deep learning MRI score proved to be the most accurate predictor of the time to conversion from MCI to AD dementia, surpassing the predictive abilities of traditional CSF biomarkers. In a cross-sectional analysis, the score showed a robust correlation with CSF tau levels and a significant but lesser correlation with CSF amyloid-beta (Aβ).¹⁰

Analysing visual attention through eye-tracking behaviour is emerging as a valuable tool for the clinical assessment of cognitive abnormalities associated with Alzheimer's disease. Utilising the capabilities of deep learning, a Multi-layered Comparison Convolution Neural Network (MC-CNN) has been developed, demonstrating the impactful role of deep learning methodologies in identifying AD. The MC-CNN has the potential to differentiate between the visual attention patterns of individuals with AD and those without cognitive impairments.¹¹

Monitoring disease progression

The Virtual Reality Functional Capacity Assessment Tool (VRFCAT), as developed by Atkins et al., was crafted to evaluate an individual's competency in performing instrumental activities in a shopping excursion. These activities encompassed tasks such as pantry organisation at home, creating a shopping list, navigating the appropriate bus route to the grocery store, conducting the shopping itself, completing the purchase transaction, and safely returning home. The study revealed that VRFCAT serves as a highly sensitive tool for assessing the functioning of instrumental activities of daily living (iADL) in individuals experiencing subjective cognitive decline.¹²

Drug discovery and development

AI has the potential to aid in drug development at various stages of the process. Machine learning (ML) algorithms can analyse different datasets such as gene expression profiles, protein-protein interaction networks, and genomic and proteomic data. By processing this information, ML algorithms could pinpoint potential therapeutic targets within the complex network of disease pathophysiology. Moreover, the use of graph neural networks (GNNs) and tree-based methods enables the elucidation of causal relationships between specific genes and diseases.¹³

Researchers developed a decision tree-based meta-classifier that utilises a network topology incorporating various biological interactions like protein-protein, metabolic, and transcription interactions, along with data on tissue expression and subcellular localisation of proteins.¹⁴ The purpose of this meta-classifier was to identify morbid genes that could potentially be targeted by drugs (druggable genes). The classifier showed promising results, accurately identifying 65% of previously known morbid genes with a precision rate of 66%, and 78% of known druggable genes with a precision of 75%. Furthermore, this tool was applied to genes not previously recognised as morbid or druggable, assigning them morbidity and druggability scores. These scores aligned with existing literature data, indicating the classifier's effectiveness in predicting the disease relevance and therapeutic potential of genes.¹⁴

AI technologies can also be utilised for screening compounds, estimating bio-activities and predicting the protein-drug interactions – which are pivotal in the hunt for efficacious drugs. Predictive models developed through AI can sieve through numerous compounds to identify those with a high probability of binding effectively to specific targets, streamlining the drug development pipeline. This technology enables more precise profile analyses and quicker rejection of non-lead structures, substantially reducing the need for extensive and expensive laboratory time.¹³

Machine learning can also be employed in the pre-clinical and clinical stages of drug development. Similarity or feature-based ML methods may be used to forecast the response of drugs on cells and the effectiveness of drug-target interactions, gauging this by binding affinity or the free energy involved in binding processes. These predictive capabilities extend to the selection of candidates for pre-clinical trials by identifying relevant human disease biomarkers and predicting potential toxic or adverse effects, thereby refining the safety profile of therapeutic candidates. By leveraging these insights, researchers can better navigate the intricate landscape of drug development, leading to more efficient and safer clinical trials.¹³

Finally, AI can aid in the approval and post-market analysis stage of drug development by using Natural Language Processing (NLP) technologies to mine the literature and prepare automated evaluations for patent applications or regulatory approval.¹³

Patient and caregiver support and education

AI is set to enhance patient care by using NLP and AI chatbots to provide easily understandable health information, answer questions, and address concerns in real time. It can encourage patient engagement through personalised experiences while connecting patients with their healthcare community. These tools can also bridge language gaps, allowing patients to receive information and support in their preferred language, thus improving access to care and adherence to treatment regimens.

Ethical considerations

The integration of AI into healthcare comes with a spectrum of ethical considerations, such as dilemmas over privacy, data security, informed consent, social disparities, and the nuances of medical interaction, including empathy and compassion. The advancement of artificial intelligence also poses a significant challenge to global social equity. As AI and automation become more prevalent, they amplify existing socioeconomic divides.¹⁵

Additionally, the accountability for errors made by AI in healthcare is a pressing concern that raises questions about whether the developers of AI or the healthcare providers employing it should be held liable. Ethical frameworks and regulatory measures are still evolving to keep pace with the rapid development of such technologies. Ensuring the safe application of LLMs and other AI technologies in patient care is imperative, necessitating a commitment to the responsible creation and implementation of AI systems.¹⁵

Could AI improve the cost burden of Alzheimer's disease?

In 2019, the estimated global healthcare cost for AD treatment reached US $1.3 trillion, encompassing a range of direct and indirect expenses.¹⁶ Research has shown that treatments capable of delaying AD-related symptom onset by just five years could lead to a 41% reduction in its prevalence and a 40% decrease in the overall cost of AD care by 2050.¹⁷ This would have a profound impact on all aspects of this disease including the patient's and caregiver's quality of life and the economic strain that this disease causes. If AI technologies can effectively improve early diagnosis and the speed of drug development, this reduction might actually be possible.

References

1. Alzheimer's Association. What is Alzheimer's disease? Accessed November 12, 2023. Available at: https://www.alz.org/alzheimers-dementia/what-is-alzheimers
2. Li X, Feng X, Sun X, Hou N, Han F, Liu Y. Global, regional, and national burden of Alzheimer's disease and other dementias, 1990-2019. Front Aging Neurosci. 2022;14:937486. Published 2022 Oct 10. doi:10.3389/fnagi.2022.937486
3. Rajan KB, Weuve J, Barnes LL, McAninch EA, Wilson RS, Evans DA. Population estimate of people with clinical Alzheimer's disease and mild cognitive impairment in the United States (2020-2060). Alzheimers Dement. 2021;17(12):1966-1975. doi:10.1002/alz.12362
4. GBD 2019 Dementia Forecasting Collaborators. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2022;7(2):e105-e125. doi:10.1016/S2468-2667(21)00249-8
5. Long C, Mclean A, Boothby A, Hollin C. Factors associated with quality of life in a cohort of forensic psychiatric in-patients. The British J. Forensic Pract. 2008;10(1):4–11. Retrieved from http://doi.org/10.1108/14636646200800002
6. Mental, physical, and speech abilities in later stages of dementia. Alzheimer's Society. Updated June 29, 2022. Accessed January 11, 2023. Retrieved from https://www.alzheimers.org.uk/about-dementia/symptoms-and-diagnosis/how-dementia-progresses/mental-and-physical-activities#:~:text=the%20later%20stages-,Mobility,be%20more%20likely%20to%20fall
7. Tabba L, Ang CS, Rose V, Siriaraya P, Stewart I, Jenkins KG, Matsangidou M. Bring the Outside In Providing Accessible Experiences Through VR for People with Dementia in Locked Psychiatric Hospitals. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI '19). 2019;236:1–15. doi:10.1145/3290605.3300466
8. Bature F, Pappas Y, Pang D, Guinn BA. Can Non-invasive Biomarkers Lead to an Earlier Diagnosis of Alzheimer's Disease?. Curr Alzheimer Res. 2021;18(11):908-913. doi:10.2174/1567205018666211207094630
9. Huseby CJ, Delvaux E, Brokaw DL, Coleman PD. Blood Transcript Biomarkers Selected by Machine Learning Algorithm Classify Neurodegenerative Diseases including Alzheimer's Disease. Biomolecules. 2022;12(11):1592. Published 2022 Oct 29. doi:10.3390/biom12111592
10. Feng X, Provenzano FA, Small SA; Alzheimer's Disease Neuroimaging Initiative. A deep learning MRI approach outperforms other biomarkers of prodromal Alzheimer's disease. Alzheimers Res Ther. 2022;14(1):45. Published 2022 Mar 29. doi:10.1186/s13195-022-00985-X
11. Zuo F, Jing P, Sun J, Duan J, Ji Y, Liu Y. Deep Learning-based Eye-Tracking Analysis for Diagnosis of Alzheimer's Disease Using 3D Comprehensive Visual Stimuli. 2023.
12. Atkins AS, Khan A, Ulshen D, et al. Assessment of Instrumental Activities of Daily Living in Older Adults with Subjective Cognitive Decline Using the Virtual Reality Functional Capacity Assessment Tool (VRFCAT). J Prev Alzheimers Dis. 2018;5(4):216-234. doi:10.14283/jpad.2018.28
13. Qureshi R, Irfan M, Gondal TM, et al. AI in drug discovery and its clinical relevance. Heliyon. 2023;9(7):e17575. doi:10.1016/j.heliyon.2023.e17575
14. Costa PR, Acencio ML, Lemke N. A machine learning approach for genome-wide prediction of morbid and druggable human genes based on systems-level data. BMC Genomics. 2010;11 Suppl 5(Suppl 5):S9. Published 2010 Dec 22. doi:10.1186/1471-2164-11-S5-S9
15. Farhud DD, Zokaei S. Ethical Issues of Artificial Intelligence in Medicine and Healthcare. Iran J Public Health. 2021;50(11):i-v. doi:10.18502/ijph.v50i11.7600
16. Wimo A, Seeher K, Cataldi R, et al. The worldwide costs of dementia in 2019. Alzheimers Dement. 2023;19(7):2865-2873. doi:10.1002/alz.12901
17. Zissimopoulos J, Crimmins E, St Clair P. The Value of Delaying Alzheimer's Disease Onset. Forum Health Econ Policy. 2014;18(1):25-39. doi:10.1515/fhep-2014-0013