How can AI help Dementia patients?

Dementia affects 50 million individuals globally. Dementia is the world's fifth leading cause of mortality. Alzheimer's disease (AD) is the most frequent cause of dementia. Mild Cognitive Impairment (MCI) is a pre-dementia stage. It is a diagnostic difficulty to distinguish Alzheimer's disease from other types of dementia and to forecast the course of persons with MCI. 

 

Biomarkers, in addition to the clinical information, aid in the differentiation of brain disease and function prior to post-mortem study. Biomarkers in dementia are characteristics that may be examined and assessed to help guide normal biological or pathological processes. 

 

They contribute to research by giving information on illness development and therapy responses, as well as identifying appropriate outcome metrics. Technology is anticipated to have used in dementia, ranging from diagnosis and assessment to care delivery and supporting the elderly in the community. 

 

Artificial intelligence (AI) is a technology that assists in the analysis, learning, and self-correction of complex data models buried in a variety of features throughout the data spectrum. Machine Learning (ML) is a subset of artificial intelligence (AI). 

 

Many scientists and academics are now employing ML algorithms and data mining methods in healthcare management to anticipate and diagnose illnesses.

 

Some of the problems that need to be answered in order to better understand the function of AI in the diagnosis and treatment of dementia are as follows:

 

• What kinds of AI applications are currently being used to diagnose dementia?

 

• How has the healthcare industry reacted to AI applications?

 

• Are there any recurring themes?

 

 

Dementia AI Applications: Introduction

 

The large bulk of AI use-cases for dementia prediction tend to fall into four broad categories:

 

Speech Monitoring

 

Machine learning is being used to analyse speech patterns in order to diagnose and track dementia development.

 

Medical Image Analysis

 

Companies are creating software that uses machine learning to assess brain degradation from scans in order to better anticipate the start of dementia.

 

Visual Indicators

 

Companies are developing methods to measure eye movement patterns in order to track and link cognitive function and brain waves.

 

Machine Learning for Genetic Data Analysis

 

Companies are utilising machine learning to examine genetic data in order to anticipate the beginning of dementia. In the article below, we present representative examples from each category, as well as the current status of each example (funds raised, pilot applications, etc.).

 

(Related blog - Machine Learning Techniques)

 

 

Speech Monitoring

 

1. Winterlight

 

Winterlight, a Canadian firm founded in 2015, promises to analyse speech patterns to identify suspected symptoms of dementia, particularly Alzheimer's disease. The business claims detection accuracy ranges from 82 to 100 per cent in clinical studies and primary care offices.

 

The Speech and Language Research model from Winterlight Lab is a software platform that uses voice recordings that are uploaded to the cloud for processing and transcription. The audio recording, which is usually a description of a well-known image, is then evaluated for 400 speech characteristics.

 

2. Canary Speech

 

Canary Speech, a Utah-based company established in 2016, claims to employ AI to aid in the study of neurological disorders through speech analysis. Initially, the firm concentrated on detecting concussions suffered by football players.

 

Algorithms are trained on audio recordings to spot minor variations in speech patterns that may suggest a concussion. The company's mobile evaluation application analyzes speech recordings and assigns a risk level score.

 

 

Image Analysis

 

1. Avalon AI

 

Avalon AI, London based, affirms the use of machine learning to detect potential signs of cognitive degenerative changes from the brain (MRI) scans.

 

Algorithms were specifically trained on 70,000 brain scan pictures obtained through collaborations with the University of Cambridge, the Donders Institute, and Imperial College. 

 

The algorithms can then spot extremely subtle abnormalities between pictures, such as aberrant alterations in anatomy, possibly assisting physicians in diagnosing dementia early.

 

Users may upload images to the company's secure network and obtain an analysis in 30 minutes using the virtualized platform. According to the company's website, there is no limit on the number of scans that can be submitted at one time.

 

(Related read - Applications of AI in healthcare)

 

Visual Indicators

 

1. Neurotrack

 

Neurotrack, a California-based company founded in 2012, claims to utilise computer vision to assess eye movement as a sign of memory health. The company's work is based on studies that say measuring the speed, direction, and patterns of eye movements might reveal vital insights into how effectively the brain functions.

 

The Imprint Memory Assessment from Neurotrack is a 5-minute exam that employs a proprietary computer vision algorithm that has been trained on data from observing eye movement patterns and is connected with cognitive performance. 

 

The system can identify patterns and provide a baseline score to test memory ability using a patient's computer camera and a sequence of eye movement activities.

 

 

Genetic Analysis

 

1. Aequa Sciences

 

Aequa Sciences, situated in London, is a member of the University of Cambridge's network, as is Avalon AI. According to the business, it analyses genetic data using machine learning and neural networks to anticipate the start of Alzheimer's disease (the most common form of dementia).

 

To discover possible signs of Alzheimer's disease, machine learning algorithms are practised on genetic data both from healthier and Alzheimer's sufferers. To better analyse and quantify the risk of acquiring Alzheimer's disease, neural networks are being created to categorise these groups into various categories.

 

 

AI Chatbots for Dementia Patients

 

AI chatbots already have the potential to benefit dementia patients and families, but the technology is still a long way from being genuinely useful and dependable. 

 

Six chatbots were included in the study, and their content and breadth were assessed using an evidence-based evaluation technique. The chatbots were evaluated primarily for their capacity to assist dementia patients and carers. 

 

The fifth and sixth chatbots were Alexa Skills voice apps, and the seventh was a text-based smartphone app. Some applications focused on teaching people about dementia, while others focused on assisting individuals with memory problems. 

 

Efficiency, performance, usefulness, affect, humanity, efficacy, contentment, and ethics and conduct were all factors considered while rating the applications.

 

• The chatbots succeeded well in terms of functionality since they were generally user-friendly. 

 

• Despite rational and easy-to-understand replies, the dialogues between people and chatbots did not look real. 

 

• Only one app passed the Turing test, which determines how well a person can distinguish between chatbot and human conversations.

 

• Chatbots have shown potential in assisting the elderly, but the current level of technology may render the gadgets unstable and difficult to use. 

 

• Furthermore, it was unclear to researchers if the chatbots were subjected to extensive empirical testing or whether the material was updated on a frequent basis. 

 

• When communicating data, the chatbots did not give references.

 

(Also read - What is Digital Health?)

 

 

The Future of Artificial Intelligence in Clinical Practice

 

While more laboratories are proving that AI is capable of effectively identifying numerous cognitive problems, even from brief samples of speech, basic ethical questions must be addressed before these technologies can be employed in medical practice. 

 

If the software is increasingly used to aid diagnosis, who will bear responsibility for wrong medical judgments if such conclusions are, in some ways, the result of statistics and the data from which they are derived? 

 

If medically important information may be retrieved from home recordings or new wearable technology, how will individual privacy be protected? Will these technologies be available just to medical practitioners, or will the general people be able to use them as well?

 

Inequalities in connectivity, potential loss of confidentiality, and unrestrained faith in innovation are significant threats that should be considered now because, if the advantages of quantifiable, reproducible, inexpensive, and demonstrates professionalism by AI outweigh the risks, a fundamental shift in health care is unavoidable.

 

Emerging AI initiatives in the battle against dementia show a wide spectrum of techniques, from speech analysis to eye movement patterns. The desire to achieve early detection and diagnosis is a frequent and fundamental element of these businesses.

 

While certain market evaluations may be fairly pricey for frequent monitoring by individuals. However, because of their simplicity of use and lower cost in contrast to evaluations, voice monitoring and picture analysis technologies retain the possibility of increasing market share.

 

Another key factor to consider when it comes to early detection is cost savings. Healthcare expenses may be decreased if individuals and affected families can prepare for the future and are not surprised by unforeseen long-term care costs.