Industries / Health / Healthcare / Home health
Nursing and therapy visits
The hands-on work done in a person's home: nursing tasks such as wound care, physical and occupational therapy, help with bathing and dressing, and reablement, which is short-term coaching that helps people relearn daily tasks so they need less help later.
6
Milestones achieved, from 1966 to 2019.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1966
Medicare pays for home health
The 1965 amendments to the US Social Security Act created Medicare, which covered hospital care, care after a hospital stay and home health care for almost all Americans aged 65 or older. It began on 1 January 1966; home health coverage was widened in 1980.
Credit
US Congress; the federal agency now called CMS
1995
Germany insures home care
Germany's long-term care insurance law was passed on 22 April 1994 and took effect on 1 January 1995. By the insurer Barmer's account, home care benefits started on 1 April 1995 and nursing-home benefits on 1 July 1996, so home care came first. The Federal Health Ministry counts 1,061,418 recipients at the end of 1995, all at home, and 1,974,197 home care recipients in 2016.
Credit
the government of Helmut Kohl, which put forward the bill
1997
A payment cut shrinks care
The US Balanced Budget Act of 1997 brought in an interim payment system, a temporary cap on what Medicare paid home health agencies. Medicare home health spending fell from $16 billion in 1996 to $7 billion in 1999; 2,571 agencies closed and 1,079 opened; and users fell from about 104 to 72 per 1,000 Medicare enrollees.
Lessonclosures explained only a small part of the drop in use, so a change in how care is paid for can shrink care more than closures do.
Credit
US Congress; measured by Korbin Liu and colleagues for the US Department of Health and Human Services
2000
Japan insures care
Japan's long-term care insurance began in 2000. Everyone aged 40 or over pays premiums, and it pays for home-visit care, such as help with bathing, toileting and cooking, as well as home-visit nursing. The health ministry's leaflet says it now supports 6.90 million service users.
Credit
Japan's Ministry of Health, Labour and Welfare
2010
Reablement measured in England
A study across 10 English councils followed people for a year after they got either up to six weeks of reablement or ordinary home care. Afterwards the reablement group's social care cost 60% less and their quality of life was better, but total cost over the year was about the same, because reablement itself cost more: £2,088 for a typical 39-day episode.
Lessonreablement improved outcomes but did not save money in its first year.
Credit
Caroline Glendinning and colleagues, Social Policy Research Unit, University of York, and PSSRU, University of Kent
2019
Nurse, therapist and handyman
In the CAPABLE trial, 300 low-income adults aged 65 and over in Baltimore received up to 10 home visits from a nurse, an occupational therapist and a handyman. Five months later they were 30% less likely than a control group to have difficulty with daily self-care (relative risk 0.70).
Lessona 2025 trial in 268 people just discharged from hospital found no significant gain, so the result depends on who receives it.
Credit
Sarah Szanton, Laura Gitlin, Bruce Leff and colleagues, Johns Hopkins
Watch for
A care robot at home
A robot nicknamed Robbie, built on Hello Robot's Stretch 4 mobile robot, is being tried in a New Hampshire home with a couple who have serious physical and cognitive disabilities. It reminds them about exercise, medicines, meals, drinking water and hygiene, and the researchers say more advanced versions can fetch items. It costs about $30,000, and the report gives no outcome data.
Achieved whena published study by someone other than the builders gives task success rates and use over several weeks in more than one home.
Claimed by
University of New Hampshire researchers, with US National Institute on Aging support, reported June 2026
Open
01
Robot arm fetches and opens
A control policy for a low-cost SO-101 robot arm that, in a staged kitchen, picks up a named object (a cup, glasses, an empty pill bottle) and places it in the open hand of a mannequin, and unscrews the plain cap of an empty bottle.
The briefCredit · unclaimed02
Home hazard check from video
A tool that takes phone video of a staged room and lists trip hazards and missing aids, such as loose rugs, no grab rail or poor lighting, tested against a list written by an occupational therapist (a professional who helps people manage daily tasks) for the same rooms. The rooms and their hazards are released as a public benchmark.
The briefCredit · unclaimed
Open 01 · the brief
Robot arm fetches and opens
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A control policy for a low-cost SO-101 robot arm that, in a staged kitchen, picks up a named object (a cup, glasses, an empty pill bottle) and places it in the open hand of a mannequin, and unscrews the plain cap of an empty bottle.
Why it is still open
The closest public dataset for this arm on Hugging Face holds 200 remote-controlled pick-and-place runs with a pill bottle and no cap opening. The best-documented in-home robot help, Carnegie Mellon's 2024 week-long trial with Henry Evans and a Stretch robot, relied on him steering the robot rather than the robot acting on its own.
Built with what exists
an SO-101 arm (about $120–150 in parts) and Hugging Face's LeRobot software (see the Note "No download"); robot models that learn from a few demonstrations, such as NVIDIA GR00T 2 (see the Note "Five demonstrations per task") or Runway Praxis-1 (see the Note "Open robot model weights"); two USB cameras.
First step
Build the arm and record 50 remote-controlled handover demonstrations with an empty bottle and a plastic cup.
Done looks like
50 filmed trials per task in a fixed mock-kitchen layout, with objects moved to new spots each time, and a published tally of successes, drops and unsafe moves. The target is at least 80% handover success and at least 60% cap removal, with video, code and data released.
Open 02 · the brief
Home hazard check from video
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A tool that takes phone video of a staged room and lists trip hazards and missing aids, such as loose rugs, no grab rail or poor lighting, tested against a list written by an occupational therapist (a professional who helps people manage daily tasks) for the same rooms. The rooms and their hazards are released as a public benchmark.
Why it is still open
The English reablement study found that quick assessment and delivery of equipment was vital and depended on getting an occupational therapist. The nearest AI work, from Columbia University and others (August 2026), ran a video AI model on simulated videos of patients at home against 12 fall risk factors, mostly about the person, and calls itself a proof of concept; it reports 85.71% accuracy on 7 simple factors and consistent failure on 2. No public set of staged rooms scored by therapists was found.
Built with what exists
a phone camera; a vision-language model that runs on a laptop through llama.cpp (see the Note "A first stable number for local AI"); 20 or more staged rooms in student flats or a training flat; the HOME FAST home hazard checklist for the therapist's reference list.
First step
Write down 10 planted-hazard setups and film each in daylight and lamplight.
Done looks like
For each of at least 20 staged rooms, filmed in two lighting conditions, the tool finds at least 85% of planted hazards with fewer than two false alarms per room, scored against two therapists' lists, with the videos and scores published.