# Pragmatic Trial of Remote tDCS and Somatosensory Training for Phantom Limb Pain with Machine Learning to Predict Treatment Response

> **NIH NIH R01** · SPAULDING REHABILITATION HOSPITAL · 2024 · $587,188

## Abstract

PROJECT SUMMARY/ABSTRACT:
Phantom limb pain (PLP) is considered an extremely hard-to-treat disorder, given that traditional treatments are
not effective in targeting the maladaptive neuronal circuits associated with chronic pain. Transcranial direct
current stimulation (tDCS) is a non-invasive, safe brain stimulation technique that has been shown to revert
maladaptive plasticity as well as reduce pain in neuropathic pain and other pain syndromes. Our previous R01
on this topic has demonstrated the efficacy of tDCS combined with somatosensory training in a controlled setting
to improve pain and that this intervention changes PLP associated cortical plasticity. Our previous R01 also
shown pain phenotypes based on PLP characteristics that are more responsive to this treatment. The objective
of this renewal is to provide novel data to address critical knowledge gaps such as (i) testing a portable device
that would reach underrepresented populations; (ii) validation of this therapy in a more pragmatic setting; (iii)
confirmation and testing of predictors of response with statistical and machine learning techniques; and (iv)
testing the parasympathetic tone changes (with the remote assessment) as a biomarker of neuropathic pain
relief. The central hypothesis is that a combination of home-based tDCS and somatosensory therapy will reduce
pain in PLP patients. Our long-term goal is to develop a cheap, efficacious, safe, and practical treatment for PLP.
Our rationale is that understanding the effects of tDCS in a real-life setting will validate this treatment for PLP
and identify predictors of response to this treatment will help health professionals better target and more precisely
treat individuals with this condition. Our specific aims will test the following hypotheses: (Aim 1) tDCS combined
with somatosensory therapy will be associated with a significantly larger effect size (of at least 1) compared to
the control condition in pain reduction; (Aim 2) identifying predictors of response of this combined treatment
using machine learning algorithms will help identify different pain phenotypes in patients with PLP and improve
their target treatment; (Aim 3) combined treatment will bolster the parasympathetic tone (as indexed by higher
heart rate variability) and reduce sympathetic activation, changes which will be correlated with PLP decreases.
This contribution is significant because, although several studies have tested the efficacy of tDCS for chronic
pain, there is a need to evaluate its effectiveness in a real-world setting, and this proposal provides critical data
to develop a safe and unique intervention to be applied at home, which can therefore increase its access to
underrepresented populations and decrease therapeutic costs. This investigation will also provide mechanistic
data on predictors of response and changes in parasympathetic tone associated with this intervention. The
proposed research is innovative because it offers a pragmatic trial d...

## Key facts

- **NIH application ID:** 10837889
- **Project number:** 5R01HD082302-08
- **Recipient organization:** SPAULDING REHABILITATION HOSPITAL
- **Principal Investigator:** Felipe Fregni
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $587,188
- **Award type:** 5
- **Project period:** 2015-08-01 → 2027-03-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10837889

## Citation

> US National Institutes of Health, RePORTER application 10837889, Pragmatic Trial of Remote tDCS and Somatosensory Training for Phantom Limb Pain with Machine Learning to Predict Treatment Response (5R01HD082302-08). Retrieved via AI Analytics 2026-05-22 from https://api.ai-analytics.org/grant/nih/10837889. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*