Screening for Alzheimer’s
DDecision support system for early identification of subjects at risk of developing dementia using machine learning models based on a range of risk factors.
Alzheimer’s Disease is one of the most common diseases in the world among neuro-degenerative diseases. Alzheimer’s is manifested in continuous and irreversible cognitive and functional deterioration. Dementia has a significant economic impact on society; According to a recent report by the American Alzheimer’s Organization, in 2021 there were over 6 million people in the United States with Alzheimer’s or other dementia diseases. The total cost of treating these patients was $ 350 billion (Alzheimer Association, 2021).
It is known that 30% of cases of dementia are preventable by balancing vascular risk factors in the years before the onset of the disease.
Early detection and prevention of at least 30% of dementia cases is about 2 million patients in the US that could receive recommendations from doctors to balance their condition with lifestyles changes that will amongst other things, reduce their cardiovascular risk factors.
Assuming that only 20% of these patients will follow these recommendations then the he total cost of treatment in the U.S. will be reduced by $ 70 billion a year.
Moreover, there are new drugs in the pipeline for the treatment of pre-clinical AD which will necessitate indemnification of at risk populations for developing AD.
In Israel there are currently 150,000 Alzheimer’s patients. Early detection is expected to contribute to a significant reduction in medical care costs and related costs to the Israeli economy. Subsequently, with the recent approval of Aducanumab for the treatment of Alzheimer’s disease, the importance of early detection of populations at risk is even more significant.
Inventors
Dr. Amir Glik, Beilinson Medical Center
Contact info
Avital Pritz, Director medical devices and digital health
For further information please contact:
avital@mor-research.comThe importance of early detection lies in the ability to prevent dementia in about 1/3 of this population by balancing cardiovascular risk factors even before the onset of the disease.
We are at the beginning of an era in which new treatments for Alzheimer's disease will be approved and whose main effect will be among populations at the beginning of the disease.
According to recent article published in the journal of internal medicine, lifestyle changes can prevent 30%-50% of Dementia cases!
The solution that we have developed is a decision support system for the early identification of subjects at risk of developing Alzheimer's disease and for measuring the progression of the disease. The system is based on machine learning models using Electronic medical records (EMR) data of Clalit Health Services patients.
The database includes personal medical information collected in the community over 20 years (and includes socio-demographic data, blood tests and other health risk factors). In this database there are about 25,000 Alzheimer's patients and about 100,000 patients in the control group. The significant size of this database allows machine learning algorithms to be trained to perform the following tasks: early detection of patients at high risk of developing Alzheimer's (screening), identification of the rate of exacerbation of the disease, and identification of patterns typical of the various stages of the disease.
Current approaches to early detection of AD at risk populations include lumbar punctures or sophisticated brain imaging techniques. These approaches are invasive and expensive.
The proposed approach uses machine learning methods on the patient's EMR and does not require the patient to perform additional tests or attend doctor's appointment.
To the best of our knowledge, no machine learning approaches have been applied to date on EMR for the early diagnosis of populations at risk for Alzheimer’s disease.
The proposed approach allows for early remote diagnosis and personalization of the treatment. It does not require the patient to undergo additional tests or see a doctor, and therefore has the potential to be more efficient and save expensive resources. There is ample evidence in the literature that early detection of at-risk populations for AD can lead to , lifestyle changes and reduction of cardiovascular risk factors which are key players in AD prevention and advanced treatment planning.
In addition, the importance of early detection of at-risk populations is growing in light of recent developments of the FDA-approved new drug for AD in early June 2021 and additional drugs that are to come. This approval is significant in light of the fact that for twenty years no new drug for AD has been approved and no drug has ever been approved as a disease modifying drug. If the drugs are indeed effective, the approach we offer will make it possible to identify populations at risk for developing AD and give them the drugs in the very early stages of the disease when its effectiveness is maximal.
A POC study done on 23,400 patients selected from the database, of whom 9,200 were diagnosed with AD dementia and 14,200 were healthy, showed promising results that indicate the ability to identify about 80% of the patients who developed Alzheimer's based on data from the ten years before the onset of the disease.
As a next step, we intend to extend this examination to additional data and test the early detection ability of the model five / ten years before the discovery of the disease.