A Spanish project is based on two widely spread tests for the detection of dementias to improve the screening of positive cases (improving the accuracy of the test) and to establish a risk profile of suffering from this disease on a large scale.
Imagine being able to detect early and accurate cognitive deterioration in any pharmacy and, from these data, to identify the risk factors of developing some type of dementia in the future. All with the comfort and flexibility of a local establishment and without having to resort, at first, to a medical center.
All this is what researchers from the Pharmacy and Mathematics departments of CEU UCH are pursuing, who together have developed two artificial intelligence algorithms that improve the screening of positive cases in the early detection of cognitive deterioration in pharmacies.
This study, published by the scientific journal Frontiers of Pharmacology, is part of the research project funded by the kNOW Alzheimer Scholarship, and has already been tested with 728 people over 65 in the first phase of this project. All of them were evaluated in pharmacies through two internationally validated tests for the detection of cognitive impairment: the Short Portable Mental State Questionnaire (SPMSQ) and the Mini-Mental State Examination (MMSE), in its Spanish version.
Through these two tests, conducted in 14 Valencian pharmacy offices, 128 cases of possible mild cognitive impairment were detected, 17.4% of the total, which were referred to primary care centers for diagnosis and subsequent referral to the neurologist. . In addition, the results of a total of 167 analysis variables were registered for early detection through these tests. Among them, factors such as age, sex, educational level, hours of sleep during the day, the habit of reading, the subjective complaint of memory loss and medication.
Where the magic lies is that these results have been subjected to a massive screening procedure, by designing two mathematical algorithms or decision trees. The first one is a discriminant decision tree, to identify the false negatives in the tests, that is, the cases of people who might suffer mild cognitive deterioration despite the test results, or also to rule out false positives. This first algorithm will allow, therefore, to improve the screening of the evaluation by means of the tests carried out in the pharmacies, to send to the doctors the positives detected in the pharmacies, for their clinical diagnosis. And also to improve the monitoring of people who, although they report symptoms of memory loss, do not obtain positive results in the tests.
The second algorithm has been designed to define patterns and design a predictive model, detecting those of the 167 variables of the evaluation by means of the two tests that are more significant for the early detection of cognitive deterioration. This predictive model is the one that allows identifying the most prominent risk factors for mild cognitive impairment.
Applied to the more than 700 cases analyzed, this predictive model has confirmed as risk factors for screening and, therefore, as more significant variables for the detection of mild cognitive impairment, the following: being a woman, sleeping more than 9 hours a day , have more than 79 years and a low reading frequency. In addition, consuming psychoanalytic, nootropic or antidepressant drugs and anti-inflammatory drugs are other of the most relevant variables detected by the algorithm. These results have confirmed the most significant variables among all those evaluated in the tests performed in pharmacies, identified as the main risk factors.