At Rise of AI 2019 in Berlin, Martin Ciupa, chief artificial intelligence officer at MindMaze, argues that AI and virtual reality in neurological healthcare, especially stroke rehabilitation, should be understood through a cybernetics lens of feedback and control loops rather than as bottom-up AI alone. He frames the problem as a growing healthcare crisis: a shrinking number of healthcare professionals per capita, rising neurological damage from increased longevity, and expensive diagnostic and therapy technology, and cites a PwC estimate that AI will contribute 15.7 trillion pounds cumulatively by 2030. He describes MindMaze's MindMotion Pro and MindMotion Go products, which use motion sensing, gamification and AI-driven recommendation engines to deliver personalized, dosage-optimized therapy to stroke patients in the sub-acute and chronic recovery phases, including a 'whack-a-mole' style game that adapts to a patient's kinematic movement. He also describes using EMG and EEG sensors for affective computing, which he calls 'artificial empathic intelligence,' to detect patient fatigue or distraction during unsupervised home therapy. He closes by tracing cybernetics back to pacemakers and brain-machine interfaces such as the Walk Again Project, and frames MindMaze's approach as a continuous, data-driven feedback loop across diagnosis, therapy and prognosis.
Key takeaways
- Ciupa frames MindMaze's approach as cybernetic feedback loops rather than purely bottom-up AI.
- He cites a PwC estimate that AI will contribute 15.7 trillion pounds cumulatively to the global economy by 2030.
- MindMotion Pro targets the short sub-acute neural plasticity window after stroke; MindMotion Go supports chronic-phase home therapy.
- He describes 'artificial empathic intelligence' using kinematic, skin-resistance and facial-recognition signals to detect patient fatigue at home.
- He credits the Walk Again Project's brain-machine-interface exoskeleton with an unexpected side effect: partial recovery of spinal connectivity in paraplegic participants.
Detailed notes
Defining the healthcare problem before the AI solution
Ciupa opens by insisting the problem be defined before AI is proposed as the solution. He cites a shrinking number of healthcare professionals per capita in the UK, rising neurological damage as longevity increases, and the high, slow-falling cost of diagnostic and therapy technology. He evaluates AI and VR solutions against four factors: productivity, diagnostic quality, personalization (recognizing individual patient patterns, such as a stroke patient who also has depression or dementia), and calls the resulting shift 'Healthcare 5.0', paralleling Industry 4.0 and 5.0.
Market opportunity and risk
He cites a PwC estimate that AI will contribute a cumulative 15.7 trillion pounds to the economy by 2030, twice the value of all gold ever mined, with healthcare cited as the second-largest use case sector for that growth, and notes VR's second-largest application sector after entertainment is also healthcare. Referencing a World Economic Forum chart, he notes that AI, biotechnology and robotics show both the greatest potential benefits and the greatest negative-consequence risk, including new IoT sensors collecting biosignal and behavioral data, so mitigation must be built into design from the start.
MindMotion: AI-guided stroke rehabilitation
Ciupa describes MindMaze's FDA-approved MindMotion platform, launching commercially in Germany, the UK and the USA. After the acute phase of a stroke (hemorrhagic or ischemic), patients enter a short sub-acute neural plasticity window where early, high-dosage, targeted therapy most improves outcomes; he references a prognosis algorithm called PREP2 using shoulder, arm and finger extension metrics, National Institute of Health data and magnetic stimulation. MindMotion Pro is used in clinics during the sub-acute phase and MindMotion Go at home during the chronic phase, using AI and gamified VR with motion-detection kinematics to keep therapy targeted and adjust as patients improve.
Affective computing and the whack-a-mole example
He describes using kinematic and physiological signals, head-drop posture, galvanic skin resistance, facial recognition and eye tracking, to detect fatigue or distraction during unattended home therapy, which he terms 'artificial empathic intelligence', paired with a virtual assistant that can suggest a break. In a 'whack-a-mole' gamification example, the system monitors elbow, shoulder and wrist movement, expands the patient's tracked workspace as they improve, and uses a level-progression recommendation engine to increase difficulty or shift focus between joints. MindMaze also uses EMG facial-muscle sensors and EEG-integrated headsets (MindMask) to detect affective and cognitive states, citing an example of detecting 'road rage' snarling versus pleasurable driving experience.
Cybernetics as the unifying feedback loop
Ciupa contrasts cybernetics, the mid-20th-century study of control and communication in man and machine, as a top-down discipline with AI's typically bottom-up, emergent approach. He traces the lineage from pacemakers to brain-machine interfaces, citing the Walk Again Project exoskeleton demonstration at the World Cup in Brazil, where paraplegic participants using a brain-machine interface unexpectedly recovered some spinal connectivity. He describes MindMaze's model as a continuous cybernetic loop of measuring input data, diagnosing health state, delivering therapy and prognosis, and feeding results back to refine neural-network models of therapy.
Quotes
There is an ongoing global health care crisis, costs increasing, access to human health practitioners is actually being stable.
Martin Ciupa
Questions answered in this talk
How does MindMaze use AI in stroke rehabilitation?
According to Martin Ciupa at Rise of AI 2019, MindMaze's MindMotion Pro and MindMotion Go platforms use AI-driven diagnosis, prognosis and therapy recommendation together with motion-sensing gamified virtual reality to deliver personalized, high-dosage therapy during the short sub-acute recovery window and to continue targeted exercises at home during the chronic phase.
What is artificial empathic intelligence?
Martin Ciupa describes at Rise of AI 2019 using kinematic data, galvanic skin resistance, facial recognition and eye tracking to detect when a patient doing unsupervised home therapy is fatigued or distracted, so a virtual assistant can respond, for example by suggesting a break.
What is cybernetics and how does it relate to AI in healthcare?
Martin Ciupa explains at Rise of AI 2019 that cybernetics is the mid-20th-century study of control and communication in man and machine, a top-down discipline he contrasts with AI's typically bottom-up approach, and that MindMaze applies it as a continuous feedback loop of measuring patient data, diagnosing, treating and refining therapy models.