# Research on the effect of TMS on insomnia patients: EEG changes and prognostic modeling

## TL;DR

This study examines how TMS affects insomnia patients' brain activity and builds a model to predict treatment outcomes.

## Contribution

The paper introduces the first SVR model using pre-treatment EEG features to predict TMS outcomes for insomnia.

## Key findings

- TMS treatment significantly changed EEG features in insomnia patients.
- An SVR model successfully predicted changes in insomnia severity and sleep quality scores.
- Key EEG features correlated with treatment outcomes were identified.

## Abstract

Insomnia (ID) is the most common clinical disorder afflicting people of all ages, races, and social classes. This study was to explore changes in the brain’s nervous system of insomnia patients after TMS treatment and construct a prognostic prediction model.

This study involved collecting EEG data of 15 patients before and after treatment, extracting features (approximate entropy, sample entropy, alignment entropy, power spectral density, median, mean, kurtosis, and skewness), and building an SVR model.

Fifteen subjects (8 females, 7 males, mean age 42 years) received 7 days of TMS on the right prefrontal lobe. Five eigenvalues were used to analyze EEG data in 5 frequency bands. Statistically significant indicator eigenvalues (p < 0.05). Paired t-test showed significant differences in PSQI and ISI total scores before and after TMS treatment, indicating its therapeutic effect. Correlation coefficients between 40 indicators and scale differences were calculated, and significant characteristic values were further analyzed. SVR models for predicting ISI and PSQI scale pre-post differences were constructed. Both had predictive ability.

This work proposes the first SVR model leveraging pre-treatment EEG features to predict TMS therapeutic outcomes for insomnia. TMS treatment can change brain waves, and the model is expected to be applied clinically, though with limitations such as small sample size and insufficiently detailed brain region division.