Thanks to techniques of machine learning, the researchers trained an algorithm capable of identifying miRNAs, in easily accessible and non-invasive blood samples, that could help predict cognitive decline prior to the development of Alzheimer’s disease. In a next stage, to reinforce the study, it will be carried out with data from patients from Peru, Colombia, Argentina and Chile.
Andrea Riquelme, Journalist.- The research of a doctor in statistics and a doctor in biomedicine, both Chileans, allowed the first stage for the development of an algorithm with the purpose of predicting the risk of developing Alzheimer’s disease in early stages. The algorithm was trained using miRNA measured in blood. miRNAs are small RNAs (ribonucleic acids) that could have the ability to regulate the expression of other genes, and are implicated in many biological processes and diseases, particularly multifactorial ones, which provides an excellent tool to investigate the mechanisms of these diseases. .
Rolando de la Cruz, academic at the Faculty of Engineering and Sciences of the Adolfo Ibáñez University and senior researcher at the Data Observatory Foundationnext to Claudia Duran-Aniotz, doctor in biomedicine, co-Director of BrainLat and academic at the UAI School of Psychology, together with the prominent neurologist Dr. Andrea Slachevsky, the neuroimaging expert Dr. Agustín Ibáñez and a multidisciplinary group, have been working for years on the development of software that integrates machine learning algorithms that serve to identify certain indicators in blood, without the need for complex and expensive neuroimaging, neuropsychology testing, and cerebrospinal fluid extraction. This software called Kit AlzMir will be a cost-effective tool to support the medical professional for the diagnostic support of Alzheimer’s disease, which is the most common form of dementia.
Currently, there are 55 million people in the world who suffer from Alzheimer’s disease and it is estimated that by the year 2050 this figure will reach 100 million. In our country there are 200,000 patients with this disease and by 2050, there will be 600,000, assuming an important health problem when considering an increasing life expectancy. Therefore, the urgency of being able to work on new disease detection mechanisms.
The pair of researchers was awarded a Fondef project two years ago and today they are participating in a Ring project, both financed by ANID, which has allowed the study to be perfected. A forthcoming investigation will use data with miRNA indicators in patients from Peru, Colombia, Argentina, and Chile, thanks to the collaboration of researchers belonging to a Latin American consortium for the study of ReDLat dementias, funded by the NIH United States. The continuation of this study aims to strengthen and enhance the software that they have been integrating new data obtained from simple blood tests, as well as the identification of other dementias. The objective is to provide a tool to support the diagnosis of the two most common types of dementia, Alzheimer’s disease and Frontotemporal dementia, and thus intervene with an early diagnosis to design treatments that can help with symptoms and improve quality of life. of the patient and their families.
In Chile, dementia screening is through a low-sensitivity EMPAM test, in addition to cerebrospinal fluid tests, which are invasive, risky, expensive, and not very accessible, like neuroimaging, therefore, “there is a critical need for the search for validated, safe, accessible, massive and cost-effective peripheral biomarkers for Alzheimer’s disease in Chile and Latin America”, indicates the researcher Claudia Duran-Aniotz.
“The revolutionary technological tool has greater sensitivity and specificity than tests used to detect the disease, but usually the diagnosis is when the disease is already developed. The use of the panel of miRNAs discovered with the machine learning algorithm to predict progressive cognitive deterioration in Alzheimer’s disease will help us to have a cost-effective and minimally invasive test to support the early diagnosis of this disease, and if we include other dementias such as Frontotemporal dementia we will have a powerful tool to support early diagnosis, since there is no cure or effective long-term treatment that can stop or reverse dementias,” said Rolando de la Cruz, doctor of statistics and author of the study.
Dr. De la Cruz explains that more than 2,000 miRNAs are obtained from each blood sample and of these, after a quality process, about 800 remain, which are used to train the machine learning algorithm. Through the algorithm we identified a panel of 7 miRNAs that could predict the risk of developing Alzheimer’s disease. With access to more data and patients from Latin America from the ReDLat cohort, we will be able to train other algorithms to predict the risk of developing other dementias in early stages such as Frontotemporal dementia. For this we not only use data from miRNAs, we also use data from other clinical, neuropsychological and neuroimaging sources, complemented by multimodal machine learning techniques. “With this advance, we seek access to early diagnostic support for some neurodegenerative diseases with a cost-effective tool without excessive expenses or invasive procedures. Knowing that these are diseases with no cure to date, technological advances today allow timely treatment and slow down the patient’s deterioration, generating an impact on their treatment and management”, concludes Dr. De la Cruz.
