Choice of machine learning models for predicting the development of psychological disorders in people with hypothireosis and hyperthireosis
The subject of this article is endocrinological diseases, namely, the analysis of complications in people with hypothyroidism and hyperthyroidism. It is known that these diseases occur asymptomatically or in a way that may indicate other possible diseases, so people do not suspect what exactly they...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Kharkiv National University of Radio Electronics
2024-06-01
|
| Series: | Сучасний стан наукових досліджень та технологій в промисловості |
| Subjects: | |
| Online Access: | https://itssi-journal.com/index.php/ittsi/article/view/486 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The subject of this article is endocrinological diseases, namely, the analysis of complications in people with hypothyroidism and hyperthyroidism. It is known that these diseases occur asymptomatically or in a way that may indicate other possible diseases, so people do not suspect what exactly they are suffering from. Later, the diseases develop to the point where complications occur in the body, some of the most dangerous of which are psychological disorders: depression, mania, aggression, etc. Therefore, the aim of this work is to develop methods for predicting the occurrence of neurological deterioration in people who have already been diagnosed with endocrinological diseases. The article solves the problem of choosing the best models for predicting the occurrence of psychological disorders in people with endocrinological problems. Machine learning methods that are widespread in the medical field were analyzed and one of them was chosen that more optimally solves all the tasks of the task. The selection of criteria took into account potential problems with medical and psychological data. The method used was linear additive convolution, which is used to select the best alternatives according to the results, with the Pareto principle, which aims to exclude less suitable alternatives because all the features have lower values than in other options. For the experiment, all features were converted into quantitative ones to calculate convolution values. The evaluation criteria are given in the paper. The following results were obtained: the forecasting model in further study of this problem will be a random forest. Conclusions: the forecasting methods were studied and a more optimal model was chosen using linear additive convolution, namely, the random forest algorithm, its advantages and disadvantages were considered. A more detailed analysis of its development will be presented in the following articles. A mathematical description of the chosen forecasting method is provided, which includes potential ways of implementation and steps for building an algorithm for one of these methods.
|
|---|---|
| ISSN: | 2522-9818 2524-2296 |