Research & Insights

Where the Evidence Is Going.

We believe rehabilitation technology should be grounded in science, shaped by clinicians, and validated by data. This page brings together the research, thought leadership, and evidence that inform how we build — and how the field is evolving.

“The next revolution is not a smarter device. It is a system that perceives, adapts, and continues learning across the full recovery journey.”

The Research Thesis

Physical AI Is the Next Frontier in Neurological Rehabilitation.

The engineering problems are largely solved. The frontier has shifted to whether Physical AI — embodied intelligence, adaptive software, and clinical oversight in one loop — can deliver recovery at home, at scale.

Platform Architecture

Two Layers. Both Required.

Most rehabilitation technology picks one lane. Heroes integrates both.

The Engineering Breakthrough

Two-way AI communication between intelligent physical devices and a digital health platform. Movement becomes data. Data drives adaptation. The device learns the person — not just the session.

The Clinical Breakthrough

An adaptive user model that builds a longitudinal clinical profile across the full recovery arc — tracking trajectory, predicting plateaus, and flagging divergence from expected recovery patterns.

The Systems Breakthrough

Neither hardware alone nor software alone closes the gap. The integration of physical devices, adaptive AI, specialist therapy programs, and continuous clinical oversight generates compounding value no isolated component produces.

Layer 1 — Physical AI

Intelligent Rehabilitation Robotics

Embedding AI directly into rehabilitation hardware — exoskeletons, wearable gloves, smart walkers.

Two-way device–platform communication. Session-by-session adaptation. Remote clinician control.

  • Physical sensors — movement, force, gait, compliance
  • Heroes AI board — personalises device to neurological profile
  • Remote adjustment — clinician-controlled between sessions
  • Independent home use — safe, governed, continuous
Research significance: first closed-loop Physical AI system for home neurorehabilitation.
Layer 2 — Clinical Intelligence

Data Integration & AI-Guided Therapy

Integrating with EMR systems and clinical workflows.

One longitudinal profile merging device data, therapy records, and patient outcomes. AI surfaces the best recommendation at every stage.

  • EMR integration — connects to existing clinical systems
  • Unified user profile — device + clinical + self-reported data
  • AI recommendations — personalised across PT, OT, ST, mental health
  • Clinician dashboards — real-time outcomes, remote programme control
Research significance: first platform merging physical AI data with clinical workflows into a single adaptive intelligence layer.

Layer 1 generates physical evidence. Layer 2 interprets it in clinical context. Together they form a closed-loop system. Neither works without the other.

Four modalities. One unified profile.

PT

Physical Therapy

Motor recovery, gait retraining, strength and balance. Closed-loop with Heroes OS device data. Adaptive to each session's performance.

OT

Occupational Therapy

Functional independence, upper limb and fine motor recovery. Informed by HandVivante and wearable device data.

ST

Speech Therapy

AI-guided therapy for post-stroke aphasia and dysarthria. Session-by-session adaptation. Clinician-reviewed throughout.

Mental Health — Integrated, Not Bolted On

Depression affects 30–50% of stroke survivors. Heroes integrates mental health as a core programme — using the same longitudinal profile to identify emotional signals early.

Research Leadership

Built by Researchers.
Not Just Builders.

Founded by people who spent careers inside the science. Full bios on the About page.

Zen Koh · Co-Founder & Co-CEO
Scale, systems & clinical governance
  • Fourier Intelligence — 57 countries, 17 research labs, USD 100M+
  • Managing Editor, JRMT
  • President, IISART · General Chair, RehabWeek 2023
  • Advisor, KITE / Toronto Rehabilitation Institute
  • Co-Founder, MotusAcademy, Zurich
Read: From Mechatronics to Managed Care
Dr. Maja Rudinac · Co-Founder & Co-CEO
Embodied AI & rehabilitation robotics
  • PhD Biorobotics, TU Delft — AI, computer vision, HRI
  • Led Interactive Robotics, Delft Robotics Institute
  • Founder, Robot Care Systems — LEA robot, 5 EU countries
  • Prince Friso Engineering Award nominee, KIVI
  • Red Dot Design Award · Shell LiveWire · Rabobank Herman Wijffels Innovation Award
  • Wired Europe 100 · Amazon Reinvent & CES main stage
Watch: Meet LEA
Michael Bankowski · Co-Founder & CEO
Clinical insight & veteran care
  • Founder, Prosperous Home Health — 10,000+ caregiving hours
  • Inventor, ROSE robot — US Patent 2021
  • SDVOSB certified · VA channel relationships
  • Former CEO, Lite Run mobility technology
  • MassRobotics 2024 alongside Spaulding & Harvard
Watch: MassRobotics 2024
ROSE · South Dakota, 2021

ROSE Robot

Built from 10,000+ hours of direct caregiving for veterans with ALS and Parkinson's.

Sit-to-stand assistance, autonomous navigation, fall prevention, medication dispensing, emergency call. Developed with SDSU, DSU, and USD. US Patent 2021.

What it proved: Physical AI for home mobility is clinically needed and technically achievable.
LEA · Delft, Netherlands, 2015

Lean Empowering Assistant — Robot Care Systems

Born from a year inside Dutch care facilities watching Parkinson's patients give up on walking.

Sensor-rich autonomous walker with AI-guided gait assistance and caregiver alerts. Deployed across 5 EU countries. Prince Friso Award nominee. LiveWIRE Award 2015.

What it proved: Embodied AI for neurorehabilitation can achieve real clinical deployment at scale.

“Two founders built the same robot on opposite sides of the world before they met. ROSE in South Dakota. LEA in Delft. Both proved the same thesis: Physical AI belongs at home.”

Research & Industry Collaboration

A World-Class Partner Network.

Global academic and industry partners at the intersection of Physical AI, mobility science, and healthcare.

Research Collaboration

Harvard Move Lab

Wearable robotics for home stroke & Parkinson's recovery

Wyss Institute

Biologically inspired engineering for rehabilitation

Harvard Medical School

Dept of Physical Medicine & Rehabilitation · Spaulding

Spaulding Rehabilitation

Harvard Medical School · BioRobotics & Motion Analysis Lab

MIT

Newman Lab for Biomechanics & Human Rehabilitation

Yale University

Faboratory · Soft robotics · wearable sensors

Industry Collaboration

COAST Autonomous

Autonomous navigation & safety for intelligent mobility hardware

MassRobotics

Boston's global robotics hub · assistive tech commercialisation

Spark Design

Human-centred design & engineering for rehabilitation hardware

OneStep

AI-powered gait analysis & clinical movement analytics

Asensei

AI-powered movement coaching & sensor-guided rehabilitation

MTEC

DoD-aligned federal health innovation · VA procurement network

Field Evidence

What the Research Currently Establishes.

Peer-reviewed studies and meta-analyses defining the current state of Physical AI and neurorehabilitation.

Systematic Review · 2025

AI in Stroke Rehabilitation: From Acute Care to Long-Term Recovery

Multiple authors · ScienceDirect / Neuroscience

AI-powered robotics provide precise, adaptive assistance across the rehabilitation continuum. AI-assisted rehabilitation demonstrably more personalised and cost-effective than conventional methods.

Read →
Meta-Analysis · 2025

Robotic Therapy in Recovery of Motor Functions After Stroke

Multiple authors · PMC / PubMed Central

Significant FMA-UE score improvements with robotic-assisted therapy across RCTs. Combination of VR and rehabilitation robots yields the strongest outcomes.

Read →
Narrative Review · 2024

Digital Health in Stroke: A Narrative Review

Silva GS, de Andrade JBC · Arquivos de Neuro-Psiquiatria

Critical gap in formal clinical validation of digital health applications despite promising results for diagnostic accuracy and treatment efficacy.

Read →
Scoping Review · 2024

Behavior Change in Digital Technology-Based Stroke Rehabilitation

Multiple authors · Journal of Medical Internet Research

Very few studies used behaviour change frameworks despite digital tools showing promise. Future development must operationalise what drives adherence.

Read →

What the evidence establishes — and where the gaps remain

  • Integrated hardware + AI + clinical oversight outperforms any isolated intervention
  • Real-world longitudinal home rehabilitation data is almost entirely absent from the literature
  • AI-assisted rehabilitation is more personalised and cost-effective than conventional methods
  • Behaviour change mechanisms in digital rehabilitation remain poorly understood
  • The majority of recovery happens at home, outside clinician observation
  • No platform has tested the full integrated Physical AI hypothesis at ecological scale

Watch & Listen

From the Field.

MassRobotics 2024 · Robotics in Age & Assistive Tech

Michael Bankowski alongside Spaulding Rehabilitation & Harvard Move Lab

Watch →

Tech Beyond Borders · Zen Koh

Neurotech, geopolitics and the next frontier of human rehabilitation

Watch →

Meet LEA · Dr. Maja Rudinac

Robot Care Systems · TU Delft · deployed in 5 EU countries

Watch →

The Paradigm Shift of Medicine to Technology · Zen Koh

DLC Talks · AI and robotics transforming rehabilitation medicine globally

Watch →

Be Part of the Evidence.

We are actively building the evidence base and looking for clinical and research partners.

Clinical Partners

Post-stroke or Parkinson's rehabilitation programs. VA networks. Neurology departments seeking real-world outcome data.

Research Institutions

Academic groups studying home rehabilitation, adherence, or behaviour change. Access to longitudinal data no single trial has generated.

Industry Partners

Hardware manufacturers, digital health platforms, and care system integrators building in the Physical AI rehabilitation space.

Join the Founding FamilyComing Soon