AI Smart Insole OEM / ODMGPS-tracking insoles, smart shoes and smart slippers for dementia care and child safety

1. Why the tracker goes in the shoe: procedural memory

The main failure mode of a wearable care device is not technical. It is that the person will not wear it. A wristband, pendant or badge is, to someone living with dementia, a strange object attached to their body. The result is that it comes off, gets hidden, or the family cannot bring themselves to insist — and monitoring ends. A device that is not being worn is not accurate; it is absent.

Footwear gets around that barrier through procedural memory. Putting shoes on to leave the house is a procedural action built over a lifetime, and it typically persists even as cognitive function declines. ASME's coverage of in-shoe monitoring devices makes exactly this argument for the shoe as carrier: people who wander still remember to put shoes on, while visible devices carry a far higher risk of removal.

The international evidence argues for realism. A feasibility study of sensor insoles in long-term care facilities found the technology technically feasible and contextually acceptable, while flagging recruitment, workflow integration, device handling and long-term adherence as real constraints. A 12-week pilot with a mean participant age of 79 reported good usability and comfort scores but comparatively low reliability scores. Both point the same way: for smart insoles, hardware reliability and encapsulation durability are the actual battleground, not the chip.

2. Scale of the need in Taiwan

This is not a niche. It is a care gap widening quickly. The Ministry of Health and Welfare's community dementia epidemiology survey puts prevalence among community-dwelling adults over 65 at 7.99% — 9.36% among women and 6.35% among men. Applied to National Development Council population projections, that is about 398,000 people over 65 living with dementia in 2026, over 470,000 by 2031 and more than 760,000 by 2046.

The wandering risk is quantified just as clearly: police statistics attribute close to 30% of missing persons aged over 65 to dementia-related wandering. Falls are more universal still. The National Health Interview Survey finds roughly one in six adults over 65 has fallen, and one in twelve has sought medical care after a fall. 52% of falls happen indoors and 48% outdoors — and the official list of risk factors explicitly includes "wearing or using unsuitable shoes or assistive devices".

What makes it urgent is the consequence. In the latest cause-of-death statistics, falls are the second leading cause of accidental injury death among adults over 65 (24.3 per 100,000), and the rate rises steeply with age.

Put the three figures together and the problem statement becomes precise: not simply knowing where someone is, but buying carers response time across three separate risk lines — wandering, falling, and gait deterioration.

3. The Long-Term Care 3.0 policy window

2026 is the most significant gear change Taiwan has made in smart assistive devices. The Executive Yuan approved the Ten-Year Long-Term Care Plan 3.0 on 31 December 2025, effective from 2026 and phased in on 1 September 2025, 1 January 2026 and 1 July 2026.

The word that matters is "rental". A smart assistive device differs commercially from a conventional one because it contains electronics, needs firmware maintenance and carries a telecoms service — an outright purchase is a high-risk decision for both families and institutions. A rental subsidy opens a subscription channel for the brand, and it means the manufacturing side has to design for repair, part replacement and refurbishment from the start, not just produce an insole that can report a coordinate.

4. Encapsulation: making rigid electronics live inside elastic foam

The real barrier is not the chip. It is heterogeneous integration. Electronics are rigid, dislike water and dislike repeated flexing. Shoe material is elastic, lives in heat and humidity, and takes thousands of loading cycles a day. Getting both into the same 3–5 mm structure without one destroying the other is what separates a working prototype from a manufacturable product.

Sensing layer: routing and strain relief

Sensing points have to land on the areas that matter in the plantar pressure map — the metatarsal heads, the lateral arch, the heel strike point — while the flexible traces connecting them must avoid the fold line where the insole bends. A trace running perpendicular to the flex axis will fatigue and crack the copper over time, so the routing deliberately detours or builds in strain-relief geometry.

Chip module: low-power positioning on two channels

The module integrates ultra-low-power positioning, eSIM 4G/LTE and Bluetooth Low Energy. The dual-channel design is a practical one: indoors or within a facility, talking to a BLE gateway saves a great deal of power, and the device only switches to cellular reporting once it leaves the site. Runtime and coverage both get served. One-touch SOS lets the wearer raise an alarm themselves while lucid.

Encapsulation: waterproofing and shock together

Perspiration is acidic and salty, and long-term ingress corrodes solder joints and traces. Potting or heat-pressed encapsulation forms a continuous water barrier that doubles as a cushion, spreading impact into the PU body instead of into the chip. Adult daily loading and children running and jumping are two entirely different stress spectra — the first is high-cycle moderate load, the second low-cycle instantaneous peaks — so encapsulation thickness and hardness have to be validated separately for each.

5. Tiered geofencing: three levels, because alert fatigue is the real enemy

Geofences usually fail by alerting too much, not too little. A single boundary has only two states, inside and outside, so ordinary activity near the edge triggers push after push — and the carer turns notifications off inside a fortnight. Tiering splits the warning into levels of intensity so that the signal that actually needs action can still be seen.

The same tiering serves institutional management. What a facility usually needs is not a real-time alarm but a retrievable track and a trend — that one resident has been moving more at night, that another's weight has been shifting persistently to one side. Those are inputs to a care plan, not emergencies.

6. Three carriers: insole, shoe and slipper

The same positioning and sensing module solves a different problem depending on what it is built into. The insole addresses going out and getting lost. The whole shoe addresses long outdoor activity. The slipper addresses the segment most often overlooked and largest in share — indoors. Recall that 52% of falls happen indoors, and at home most older people wear no shoes at all. That is the gap neither an insole nor a shoe can reach.

Smart shoe: move the module out of the flex zone

The big technical advantage of embedding into a whole shoe is that the electronics can sit where the shoe does not bend. An insole is limited by thickness and by forefoot flex, so the module usually has to squeeze into a narrow area at the arch or heel. A whole shoe lets it be buried in the midsole or heel wall, where deformation is far lower than at the forefoot — encapsulation fatigue life goes up substantially and battery capacity no longer has to compromise for thickness.

The cost is adoption friction. The user has to accept a new pair of shoes, with all the size, last and appearance preferences that implies, and when the shoe wears out the module usually goes with it. That makes the smart shoe a better fit for settings where footwear is issued centrally — occupational safety, special education campuses, institutional procurement — than for retail.

Because we hold both the sole tooling and the shoe material capability, the cavity for the module can be designed into the midsole at the tooling stage rather than cut into a finished part afterwards. Where slip or safety requirements apply, see CNS Certified Safety Shoe OEM and All-Material Outsole Manufacturing.

Smart slipper: the indoor 52%

A slipper is the only footwear an older person naturally puts on at home, which makes it the logical carrier for indoor sensing. Indoor falls are 52% of the total, and getting up at night for the bathroom is the high-risk window inside that: poor light, postural hypotension on waking, and a slipper whose front edge is too high or whose sole is too slick — a textbook instance of the official risk factor "wearing or using unsuitable shoes or assistive devices".

So the design centre of gravity moves. The sensing objective shifts from positioning to fall detection and bed-exit alerting:

• Structure before electronics — a low front edge so it cannot catch, heel coverage so it cannot fall off, and a tread matched to the tile and wood floors found in homes
• Bed-exit detection — pressure sensing establishes that the person has stood up, which can trigger bedside lighting or a carer prompt at night
• Fall inference — acceleration and plantar pressure disappearing together as the suspected-fall signal, which is more robust than either sensor alone
• Dock charging — the advantage unique to slippers. A slipper has a fixed resting place, so it can charge automatically when put back, which sidesteps the hardest practical problem in this whole category: remembering to charge the device

We have long produced slipper tooling and one-piece slipper manufacturing, so the slipper body, the anti-slip outsole and the module cavity can all be planned within a single tool.

Choosing between them

7. Long-term care applications: from position to gait trend

What makes an embedded insole or slipper valuable is that it is a sensor as well as a tracker. Beyond location, the flexible pressure layer accumulates three kinds of behavioural data:

For an institution, the point of this data is turning after-the-fact review into before-the-fact intervention. Health promotion statistics already identify indoors as the primary location of falls (52%), with indoor injury locations ranking as living room, bedroom, bathroom, kitchen or dining room, balcony and stairs. Once positioning resolves to room level and gait stability is tracked alongside it, fall prevention can move upstream from incident reporting to circulation and environment modification.

8. Children and special education: same technology, different stress spectrum

Shrinking the adult version to a child's size is the most common development error here. Children and older adults are different design problems on three axes.

Foot structure: the paediatric growth curve

A child's arch is still developing, and foot proportions and arch height differ from an adult's. The insole body has to be tooled to the paediatric foot curve rather than scaled proportionally, and sensing point locations move with the different pressure distribution.

Stress conditions: much higher peaks

Children generate high instantaneous impact peaks in varied directions, but fewer total cycles than an adult's daily commute walking. Encapsulation therefore has to prioritise impact resistance over fatigue endurance. We validate the two stress spectra separately rather than judging children's footwear reliability by adult test conditions.

Geofence logic: the safe zone centre moves

For children the geofence usually centres on a kindergarten, primary school or after-school centre, and it needs to switch dynamically around the school run. Once the child leaves the defined area, the system notifies parents, the school and the care provider — compressing "search afterwards" into "know immediately". The same architecture serves special education units, giving students prone to wandering protection without a visible label.

9. Manufacturing reliability: from "it works" to "it works for three years"

A prototype lighting up and a production unit surviving perspiration, loading, washing and knocks for an entire rental cycle are different orders of difficulty. A large part of why the facility feasibility study named "device handling" and "adherence" as adoption constraints is that hardware wears out faster in the real world than in the lab.

• Waterproofing — encapsulation continuity and solder joint corrosion under long-term perspiration ingress
• Shock — adult high-cycle loading and child high-peak impact, validated separately
• Flex — fatigue cracking of flexible traces at the insole fold line
• Serviceability — replaceable parts and refurbishment, to match the rental subsidy business model

The fourth deserves particular attention from brand owners. Under a rental model an insole's life does not end at the sale — it has to come back, be cleaned, have its battery replaced and be rented again. That forces demountable encapsulation rather than fully potted construction, and it shapes material choice and joining method. It is a problem conventional single-use insole manufacturing never encounters, and it is where two decades of tooling and material work actually pays off: we do not put electronics into an insole, we start from the structure and service life of the insole and work backwards to where the electronics can go.

One thing does not change with the addition of AI: comfort. If the wearer will not keep it on because it feels like a foreign object, every specification above is worthless. That is why the base of the smart insole remains our high-density PU functional insole structure rather than a support geometry compromised to make room for components.

10. Frequently asked