Industries / Health / Healthcare / Physical therapy
Hands-on treatment
Hands-on treatment is what a therapist does to a patient's body with hands or a machine: massage, moving and manipulating joints, and passive treatments such as electrical stimulation, therapeutic ultrasound, heat, traction and taping.
7
Milestones achieved, from 1965 to 2019.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1965
Gentle oscillations before thrusts
Geoff Maitland, an Australian physiotherapist, studied with doctors, osteopaths, chiropractors and bonesetters abroad and was invited to teach in Britain in 1965. He taught gentle, repeated oscillating movements of a joint to find and work toward its limit of motion, and held that these mobilizations were often better than a fast thrust; with Jenny Hickling he added movement diagrams to describe where a joint stiffens or hurts. By Erl Pettman's history, these gentle mobilizations became a core part of manual therapy training in Britain and around the world.
Credit
Geoff Maitland; Jenny Hickling
1967
Electric pulses ease pain
Patrick Wall and William Sweet stimulated the sensory nerves supplying painful areas in eight patients with 0.1-millisecond pulses at 100 per second, turned up until the patient felt tingling; during stimulation, pressing on the sore spots no longer hurt, and four patients had relief for more than half an hour after 2 minutes of stimulation. Their work followed the gate control theory that Ronald Melzack and Wall published in 1965, which held that signals from touch nerves can close a "gate" on pain signals in the spinal cord. A 2019 review says this model led to TENS (transcutaneous electrical nerve stimulation), which passes current through the skin and which patients can use on themselves.
Credit
Patrick D. Wall and William H. Sweet; Ronald Melzack
2001
Ultrasound no better than placebo
Therapeutic ultrasound passes sound waves too high to hear into tissue, and after more than 60 years it was one of the most used machines in physical therapy. Valma Robertson and Kerry Baker reviewed 35 randomized trials from 1975 to 1999 that compared real with sham ultrasound; of the 10 with acceptable methods, 8 found no difference and 2 found a benefit, for carpal tunnel syndrome and calcium deposits in the shoulder. Doses varied widely "for no discernable reason".
Lessona treatment can stay in daily use for decades without trials that compare it with a fake version.
Credit
Valma J. Robertson and Kerry G. Baker (Physical Therapy, 2001)
2008
Stroke risk traced to symptoms
A fast thrust to the neck has been blamed for tears in the arteries at the back of the neck that can cause a stroke. A study of 818 such strokes in Ontario hospitals from 1993 to 2002, matched against health billing records, found that stroke patients were more likely than matched controls to have seen a chiropractor in the year before, and to the same degree more likely to have seen a family doctor; the authors concluded the visits were probably made for the neck pain and headache of an artery tear already under way. The world body of manual physical therapists, IFOMPT, approved a framework for screening the neck for artery problems before treatment in October 2012, with 22 member countries, and revised it in 2022.
Lessona rare harm seen after a treatment may come from the condition that brought the patient in, so careful screening matters more than the technique alone.
Credit
J. David Cassidy and colleagues (Spine, 2008); Alison Rushton and colleagues for IFOMPT
2014
Tape no better than sham
Kinesio Taping uses stretchy tape stuck to the skin over sore muscles and joints. A review of 12 randomized trials with 495 people with shoulder, knee, back, neck and heel pain found it no better than sham taping or other treatments; where it did better, the effects were small or the trials were of low quality.
Credit
P. do C. Parreira, L. O. Costa and colleagues (Journal of Physiotherapy, 2014)
2016
Passive machines dropped for backs
England's guideline on low back pain and sciatica, from NICE (the National Institute for Health and Care Excellence), tells clinicians not to offer traction, ultrasound, TENS, interferential therapy or acupuncture for low back pain. It allows manual therapy (manipulation, mobilization or massage) only as part of a treatment package that includes exercise.
Lessononce trials showed passive treatments did little on their own, the advice moved to hands-on care as a support for exercise.
Credit
NICE, guideline NG59, published 30 November 2016
2019
Manipulation equals other care
A review of 47 randomized trials with 9,211 adults with chronic low back pain found that spinal manipulation or mobilization relieved pain about as much as other recommended treatments, such as exercise, with a small extra gain in function.
Credit
Sidney Rubinstein, Annemarie de Zoete, Maurits van Tulder and colleagues, Vrije Universiteit Amsterdam (BMJ, 2019)
Watch for
Robot massage in spas
Aescape's robot massage, named one of TIME's Best Inventions of 2024, uses robotics and AI to adapt to the user's body contours and chosen pressure while the user stays clothed; in a self-guided session the user sets pressure, target muscles and recovery goals on a touchscreen. Massage Envy, which calls itself the largest US provider of massage across its franchise network, said select franchised locations would begin offering it.
Achieved whenan outside laboratory or trial publishes measurements of the force the robot delivers against the setting chosen, or compares its effect on pain or stiffness with a human massage therapist's.
Claimed by
Massage Envy and Aescape, December 2025
Robot holds a steady force
Hondak Robotics of Huizhou, China, says its massage robot finds the body's contours with three-dimensional vision and uses force and torque sensing so that it can "maintain a stable force as it works over different parts of the body rather than simply moving along a predetermined path". It switches between treatment heads for pressing, kneading, heating, cupping and shock-wave massage.
Achieved whenindependent force measurements on a standard test body, or a randomized trial in patients, are published.
Claimed by
Hondak Robotics, in China Daily, October 2026
Open
01
Count passive treatments in Medicare
An open yearly series, 2013 to 2024, of Medicare Part B units and payments for passive treatments, such as therapeutic ultrasound (CPT code 97035), unattended electrical stimulation (G0283) and mechanical traction (97012), set beside active treatment such as therapeutic exercise (97110). Break it down by state and by type of practitioner, show how widely the passive share varies between practices, and mark the dates of guidelines that advised against passive treatments, such as NICE NG59 in 2016.
The briefCredit · unclaimed02
Measure robot massage pressure
A public test of how much force massage robots actually apply compared with the setting chosen. Lay a thin pressure-mapping mat over a standard test body (a silicone back on a force plate), run each robot through the same routines at low, medium and high settings, and have licensed massage therapists perform the same strokes on the same test body for comparison. Repeat on healthy adult volunteers using robots that are open to the public.
The briefCredit · unclaimed
Open 01 · the brief
Count passive treatments in Medicare
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open yearly series, 2013 to 2024, of Medicare Part B units and payments for passive treatments, such as therapeutic ultrasound (CPT code 97035), unattended electrical stimulation (G0283) and mechanical traction (97012), set beside active treatment such as therapeutic exercise (97110). Break it down by state and by type of practitioner, show how widely the passive share varies between practices, and mark the dates of guidelines that advised against passive treatments, such as NICE NG59 in 2016.
Why it is still open
CMS's public files show the trend can be measured: Medicare paid for 3,168,981 office units of therapeutic ultrasound in 2013 and 1,063,938 in 2024. But a 2023 scoping review of 33 studies of low-value physiotherapy for low back pain found that 67% drew their data from surveys, most looked only at the provider's view, and none took society's. The nearest claims work is a one-off comparative billing report from Palmetto GBA, a Medicare contractor, on about 15,000 physical therapists' claims from July 2015 to June 2016, sent to the therapists themselves; no public, repeatable series built from claims was found.
Built with what exists
CMS's Medicare Physician & Other Practitioners files, by geography and service and by provider and service, for 2013 to 2024, through the data.cms.gov API; Python with pandas or DuckDB; GitHub for code and tables.
First step
Pull the national rows for 97035, G0283, 97012 and 97110 from the 2013 and 2024 by-geography files and plot the change.
Done looks like
Units and payments for at least five passive and three active codes, for 12 years and every state, split by practitioner type; national totals that match CMS's files exactly; the distribution of the passive share across practices billing at least 1,000 therapy units a year; code and tables published and rerun within a month of CMS releasing 2025 data.
Open 02 · the brief
Measure robot massage pressure
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test of how much force massage robots actually apply compared with the setting chosen. Lay a thin pressure-mapping mat over a standard test body (a silicone back on a force plate), run each robot through the same routines at low, medium and high settings, and have licensed massage therapists perform the same strokes on the same test body for comparison. Repeat on healthy adult volunteers using robots that are open to the public.
Why it is still open
A 2024 systematic review found 17 studies of robot massage, 15 of them preliminary trials and only one randomized trial. Research robots report their own force tracking, and the makers of robots now offered to the public publish no outside force measurements (see Watch for).
Built with what exists
a pressure-mapping mat (such as Tekscan's or XSENSOR's) or a low-cost array of force-sensing resistors read by an Arduino; a force plate; a cast silicone back; paid sessions on robots open to the public, such as Aescape's; Python for analysis.
First step
Cast the silicone back and record five therapists' strokes on it at what each calls light, medium and firm pressure.
Done looks like
For at least two robots and five therapists, peak and mean pressure in kilopascals at each setting over 10 runs, the run-to-run spread, and the gap between set and delivered force; at least 20 volunteer sessions; protocol, data and code published.