HHS TOPx Tech Sprint · AI & Invisible Illness
See the Health Spaces MVP.
Our final showcase for the HHS TOPx Tech Sprint. Brittany Martin, our president, walks through Nora’s Health Space. Nora is a fictional patient, and the demo uses fictional data.
Our final submission
The information between appointments, finally held together.
Chrysalis Health Spaces is a patient advocacy and care management tool for people with complex chronic and invisible illness, including those seeking a diagnosis and caregivers supporting them.
The problem we’re solving
Much of the information that matters was never captured in the first place.
A person living with complex chronic illness can see multiple specialists, undergo years of testing, try dozens of interventions, and still have no coherent account of what has happened to them.
Records are organized around encounters, providers, tests and diagnoses. They rarely capture what life looked like before symptoms began, which activities became impossible, why treatments were discontinued, or how one decision affected everything that came after.
The patient becomes the only person connecting those pieces, often from memory, during an appointment, while living with fatigue, brain fog or pain.
“Every time I went to a doctor, it was like starting over from scratch.”A research participant
How Health Spaces works
It starts wherever the patient can start.
Built on Chrysalis’s existing personalized AI and persistent-memory infrastructure. Rather than asking an exhausted patient to be a data clerk, it gives them a private, conversational workspace to reconstruct their history and keep adding to it as life happens.
Unexplained symptoms
May begin with what’s happening today.
A long history
May need to reconstruct a history that stretches back to childhood.
Helping someone else
May be helping someone who can no longer reliably remember or communicate their experiences.
Incomplete histories, approximate dates and information that emerges gradually all fit. No one has to remember everything in one sitting.
The course of a person’s health over time
Story Anchors and the timeline aren’t limited to diagnoses and medical events. Changes in function, employment, caregiving, living circumstances, support and access to care can all matter to understanding what happened and when.
Illness, in the patient’s own terms
Two people with the same diagnosis may experience it very differently. Symptom and Condition Expression Cards preserve those differences rather than reduce someone’s experience to a diagnosis code or symptom score. Conditions are labeled clinician diagnosed, self diagnosed or candidate.
What actually happened
Medications and supplements, and also changes in activity, diet, routines and accommodations. A treatment can help one problem while worsening another, or become impossible to continue because of cost, access, or tolerability.
Focused follow-up questions, not daily forms
Rather than an extensive daily form, Chrysalis asks focused follow-up questions informed by what the person has already shared. Change over time stays visible without making exhaustive self-monitoring a requirement.
Context, alongside what patients report
Check-ins line up over time with weather and air quality, and wearables like Apple Watch are coming next. Events occurring together don’t establish medical causation, so the record keeps each source separate.
From personal history to an appointment
Two kinds of handoff.
A new specialist needs the larger picture.
A baseline handoff helps a practitioner understand a patient’s larger history, current priorities, functional limitations, prior treatment experiences, and unresolved concerns.
Someone following along needs what’s changed.
An interval handoff focuses on what has changed since the last visit: new concerns, changes in existing symptoms, treatments attempted or discontinued, shifts in function, and questions requiring attention.
AI can help organize and summarize, but it’s not the authority on which experiences are true, which concerns matter, or what medical conclusions should follow.
A readable, downloadable handoff brings patient-held context into an existing visit. More integrated exchange follows on our interoperability roadmap.
Diagnosis, treatment decisions and clinical interpretation remain with qualified professionals.
Capturing what conventional health data misses
A patient-centered account asks different questions.
We mapped our patient-context model against all 626 variables in the 2024 National Health Interview Survey Sample Adult questionnaire. Eight areas had no corresponding question, including:
That doesn’t establish that these experiences are missing from every health dataset. We’re also examining the NINDS ME/CFS Common Data Elements and the DePaul Symptom Questionnaire-2, without administering them or claiming patient narratives produce validated scores.
Four federal data resources, connected to the patient’s account
Every entry keeps its source, so AI can tell who said what:
Events occurring together don’t establish medical causation. Patients and practitioners can examine possible relationships without being presented with conclusions the evidence can’t support.
Designing for the people most likely to be left out
The hardest part may be getting the information captured at all.
A tool can have excellent technical capabilities and still fail the people it was built for if it requires too much attention, consistency, physical effort, or cognitive energy.
Missing isn’t nothing
Someone with post-exertional malaise may be least able to document their experience when symptoms are most severe. Missing observations aren’t evidence that nothing happened.
Stable isn’t always better
A person who has adapted by eliminating activities may look more stable on some measures while their independence and overall function have declined.
- Accepts partial information
- Preserves uncertainty
- Lets people return without starting over
- Still useful without regular tracking
“Almost like a cheat sheet to treating me, for a doctor.”A research participant
Why Chrysalis can deliver this
Extending a platform we’ve already built.
Health Spaces applies Chrysalis’s existing infrastructure for personalized, portable AI context to complex chronic and invisible illness, rather than proposing an architecture that depends on future funding or research breakthroughs.
Brent Martin
Founder and Systems ArchitectMore than 25 years of lived experience with chronic illness, with systems engineering, executive leadership and FDA/USDA regulatory compliance. He created Metabolic Me in 2009 and co-founded Lyme Less Live More in 2013.
Brittany Martin
PresidentLeads operations, strategy, partnerships and commercialization, with a background in executive leadership, organizational development and community-building.
Our engineering team
AI and data platformsTwo decades architecting AI and data platforms at Capital One, PayPal, eBay, Ancestry and PwC, including systems serving 2.5 million public servants.
Where this can lead
First, genuinely useful to the person living with illness.
- The MVPBaseline history, updates and patient-reviewed handoffs
- SOC 2 and HIPAAAudit expected to complete
- Free beta groupsRecruited through patient and practitioner networks
- Broader accessAfter privacy, onboarding and usefulness testing
- EHR interoperabilityThrough HL7 FHIR and USCDI
With separate consent, research governance and validation, aggregated histories could help researchers and agencies examine diagnostic journeys, treatment access, costs, and effects on work, independence and caregiving.
The information that has always existed between appointments deserves a place in the systems we use to understand illness.
Be one of the first to open your case.
The beta is free and by invitation. We'll invite people in small groups, then open Chrysalis in cohorts in early 2027.
Chrysalis was a Phase 1 winner in a federal challenge. That is not an endorsement by HHS, NIH or FDA. The demo and the screens on this page use fictional patient data. Evidence so far is qualitative discovery and beta feedback, not measured outcomes.