ANM
ThesisFrameworkOpportunitiesLibraryField Notes
한국어Admin
AI-Native Medicine
Opportunity Letter한국어
All issues

ISSUE 006 · Oct 8, 2026 · 8 min read

What if neuromodulation adapted when gait worsened before pain scores did?

For chemotherapy-induced peripheral neuropathy, combine app-controlled TENS with gait sensing and symptom diaries to test a 12-week recovery episode that learns individual response instead of delivering a fixed dose.
The issue in one sentence

For chemotherapy-induced peripheral neuropathy, combine app-controlled TENS with gait sensing and symptom diaries to test a 12-week recovery episode that learns individual response instead of delivering a fixed dose.

Diagram 01

A stimulation loop informed by symptoms and gait

01Symptom diary→
02Gait · balance→
03Signal-quality gate→
04Response phenotype→
05Clinician review→
06Bounded adjustment→
07Function · falls

ANM proposed model. Stimulation changes only weekly within pre-authorized bounds; no single sensor signal autonomously changes treatment.

01Clinical problem · Evidence

A pain score can leave numbness, balance loss, and fall risk outside the treatment plan.

CIPN after taxane, platinum, and other neurotoxic chemotherapy can leave pain, burning, tingling, numbness, and impaired walking. Symptoms may persist for months, while pain and function do not necessarily move together and options for non-painful symptoms are especially limited.

02Technology stack · Proposal

Connect wearable TENS to brief gait tests and patient-reported outcomes.

Place app-controlled noninvasive TENS, phone or inertial-sensor 30-second stance and Timed Up and Go tests, and daily pain, cramping, and numbness reports on one timeline. AI classifies response phenotypes and proposes the next week’s plan for clinician review rather than diagnosing.

03Research question · Hypothesis

Does symptom-and-gait adaptation improve function beyond fixed stimulation?

Test hierarchically whether weekly, bounded sensor-informed TENS improves the 12-week EORTC QLQ-CIPN20 sensory score and a digital gait composite versus the same active device on a fixed protocol, while also estimating active treatment effects against sham.

04First study · Proposal

Twelve cancer centers, 180 survivors, three arms, 12 weeks.

Randomize adults with moderate or worse chronic CIPN at least three months after taxane or platinum therapy 1:1:1 to identical-appearing sham, fixed active TENS, or sensor-informed active TENS. Co-primary outcomes are blinded 12-week change in the EORTC QLQ-CIPN20 sensory score and a prespecified gait-balance composite. Daily pain, cramping and numbness, real-world falls and near-falls, gait speed, adherence, skin events, and discontinuation are secondary outcomes.

05Defensible data · Hypothesis

The moat is not device uptime—it is who responded, and in which functional domain.

Link chemotherapy class and cumulative dose, neuropathy phenotype, stimulation waveform and exposure, symptom trajectories, gait and balance, falls, rehabilitation referrals, and clinician decisions. Separating pain responders from mobility responders can refine eligibility and dosing in subsequent trials.

06Venture thesis · Assumption

Sell a 12-week functional-recovery episode, not another stimulator.

Initial customers are cancer-center survivorship clinics and sponsors of supportive-care trials. Bundle device rental, remote setup, gait assessment, alert review, rehabilitation referral, and outcomes reporting per patient episode. Test payer outcomes contracts only after prospective evidence shows better symptoms or function and less staff burden.

Evidence map

Feasibility exists; personalization remains unproven

142 people

Randomized in the six-site wearable-TENS proof-of-concept trial

130 · 91%

Participants completing the six-week assessment and completion rate

No observed effect

No observed benefit for tingling or numbness in that trial

425 people

Cancer survivors assessed for neuropathy- and fall-related gait domains with inertial sensors

Evidence

Established evidence versus the untested venture thesis

Established: a six-site randomized trial published in 2024 demonstrated multicenter feasibility and 91% completion for app-controlled wearable TENS. It showed preliminary signals for pain and cramping but no observed effect on tingling or numbness. Because the trial prespecified alpha at 0.20 for preliminary efficacy, it was not confirmatory. In a separate observational analysis of 425 female cancer survivors, CIPN, prior falls, and prospective falls related to different sensor-derived gait and balance domains, but the study did not predict treatment response. A phase II trial of Scrambler therapy plus duloxetine remains active in 2026. Evidence that sensor-guided personalization beats fixed TENS, prevents falls, or supports a purchasable 12-week episode does not yet exist; those remain research and venture hypotheses.

Wearable-TENS randomized trial Gait and fall-sensor study Active Scrambler phase II trial FDA TENS-device record

Failure conditions and safety, ethics, and regulatory risks

Clinical failure condition

Drop the personalization claim unless the sensor-informed arm exceeds fixed active TENS by at least five EORTC sensory points or the prespecified minimum effect on the gait composite.

Pain-only improvement

Do not market functional recovery if pain improves but numbness, balance, or falls do not. Report benefit and deterioration separately by symptom domain.

Sensor error and over-adjustment

Missing wear time, an absent phone, or mobility-aid use can mimic decline. Never adjust from a single reading; require repeat confirmation and clinician authorization.

Red flags that must not be masked

Abrupt asymmetric weakness, bowel or bladder change, fever, or new back pain is not presumed to be CIPN; route urgently for evaluation of metastasis, cord compression, infection, or other causes.

Stimulation safety and access

Monitor skin injury, discomfort, falls, and interference with implanted electrical devices; exclude pregnancy, severe skin disease, and implanted life-support devices. Report attrition by smartphone literacy and connectivity.

Regulatory and business boundary

A 510(k) for general transcutaneous pain relief does not clear CIPN personalization or autonomous parameter changes. A new indication or algorithm may require separate device review; reject the episode model if adoption stays below 70% or clinician workload rises.

Bibliography

References

Established facts are separated from research and venture hypotheses that still require validation.

  1. 01

    Gewandter JS, Culakova E, Davis JN, et al. Wireless Transcutaneous Electrical Nerve Stimulation for Chronic Chemotherapy-Induced Peripheral Neuropathy: A Proof-of-Concept Randomized Clinical Trial. J Pain. 2024;25(5):104431.

    Randomized 142 participants across six sites; 130 completed. Preliminary signals favored pain and cramping, not tingling or numbness, and the trial was not confirmatory.PMID 37993030 · NCT04367480
  2. 02

    Shah VV, Muzyka D, Guidarelli C, et al. Chemotherapy-Induced Peripheral Neuropathy and Falls in Cancer Survivors Relate to Digital Balance and Gait Impairments. JCO Precis Oncol. 2024;8:e2300312.

    Used six inertial sensors in 425 female cancer survivors. Different domains related to CIPN, prior falls, and prospective falls, but the analysis was observational.PMID 38885463 · DOI 10.1200/PO.23.00312
  3. 03

    National Cancer Institute. Scrambler Therapy with Duloxetine-Based Usual Care versus Duloxetine-Based Usual Care Alone for Chemotherapy-Induced Peripheral Neuropathy.

    An active phase II trial adding ten Scrambler sessions to duloxetine-based usual care, with 50% pain reduction at day 35 as the primary objective.NCT06914557 · Phase II · Active 2026
  4. 04

    U.S. Food and Drug Administration. Scrambler Therapy Technology, Model ST-5A. 510(k) K201458.

    A 510(k) record for a transcutaneous nerve stimulator for pain relief—not clearance of sensor-guided personalization for CIPN.K201458 · Dec 23, 2020
The next issue publishes every day at 9 AM KST.All issues