# Prediction of treatment responsiveness to home-based transcranial photobiomodulation (tPBM) intervention for cognitive decline using fNIRS concurrently recorded during tPBM

## TL;DR

This study shows how brain activity data collected during a light therapy can predict if someone will benefit from the treatment for cognitive decline.

## Contribution

A novel method using fNIRS and graph-theoretical analysis to predict treatment response to tPBM before therapy completion.

## Key findings

- Graph-theoretical indices from fNIRS data correlated with cognitive improvement during tPBM.
- Early prediction of non-responders was possible using regression thresholds from initial treatment sessions.
- 13 out of 29 participants were correctly identified as non-responders before treatment completion.

## Abstract

Transcranial Photobiomodulation (tPBM) has attracted growing interest as an intervention to mitigate cognitive decline in older adults. However, some individuals do not respond to tPBM. This study explored the feasibility of predicting treatment responsiveness using functional near-infrared spectroscopy (fNIRS) recorded during therapy with a device integrating tPBM and fNIRS.

Twenty-nine participants with cognitive decline underwent 12-week home-based tPBM intervention with concurrent fNIRS acquisition. Notably, fNIRS data were collected using the existing tPBM light sources, without additional hardware. After termination of the intervention, patients were classified as responders or non-responders based on changes in the global cognitive score (ΔGCS), which reflects multiple cognitive domains. Fourteen participants were classified as responders and 15 as non-responders. fNIRS data from the initial 15 trials were segmented into 5 periods. Linear regression analysis was performed to evaluate the changes in graph-theoretical indices calculated from the functional connectivity analysis of fNIRS and their relationship with ΔGCS. Participants with regression values below a designated threshold were predicted as non-responders.

Significant negative correlations between ΔGCS and the changes in graph-theoretical indices were observed in periods 3–5. Participants with regression values below a designated threshold were predicted as non-responders. In total, 13 participants were identified as non-responders, with 11 confirmed as non-responders after tPBM therapy.

We explored the feasibility of applying graph-theoretical network analysis to fNIRS data for the early identification of non-responders to tPBM treatment before its completion. This novel approach can potentially enhance treatment efficacy by allowing for timely treatment planning.