Create Jira Ticket
This idea cannot be sent to Jira.
WholeContext AI™ — Connected Clinical Intelligence for Earlier Diagnosis, Treatment Optimization, and Coordinated Care
WholeContext AI™ — Connected Clinical Intelligence for Earlier Diagnosis, Treatment Optimization, and Coordinated Care
WholeContext AI connects fragmented medical information, daily symptoms, changes in function and relevant U.S. public data into a patient-reviewed history that supports investigation, treatment and recovery. Beginning with Lyme disease and overlapping invisible illnesses, it addresses a familiar breakdown: patients and families repeatedly reconstruct the story while clinicians must make sense of incomplete information under time pressure.
I’m Kimberly Chulis, PhD, founder of Core Analytics and creator of WholeContext AI. My experience includes serving as a Principal Data Science Manager at Microsoft, teaching graduate data science at Northwestern University and doctoral research at Purdue examining health-related social-media data. Helping my mother navigate appointments, results and changing symptoms made the problem personal. I lead the product, architecture, research and AI-assisted development, incorporating patient, caregiver, physician and technical feedback.
That feedback shaped concrete choices: patients can review and correct their histories, control sharing, preserve the source of an observation and define recovery through goals that matter in daily life.
From fragmented information to a care conversation
Before a visit, patients bring together dated records, journals, symptoms, pain, treatments and daily function. Caregiver observations retain their attribution. Selected evidence becomes a concise, source-linked clinician packet.
During review, WholeContext Digital Twin™ lets clinicians explore the dated history, inspect supporting and conflicting evidence, identify missing information and examine recorded symptoms and function alongside interventions. Historical views preserve what was known at an earlier point; hypothetical scenarios remain separate from the patient’s record.
Between visits, patients continue tracking symptoms, function and treatment observations, creating a basis for reassessment.
In Marion’s fictional journey, recovery has a concrete meaning: dressing and reaching comfortably. The Twin connects that goal with dated observations and interventions so a clinician can investigate what changed, when it changed and what warrants follow-up.
Making public data relevant to the individual
WholeContext Discovery Lab™ and OpenDataDelta let reviewers compare a patient’s evidence with and without selected public-data context. Stored CDC, HRSA and Census-derived information can introduce county Lyme reporting, physician supply and transportation barriers into questions about exposure and feasible care. Source periods, geography and limitations remain visible; population statistics do not become an individual diagnosis.
The research program catalogs 171 public-data sources and candidate resources. The MVP uses selected stored sources, with broader integration organized through the accompanying data matrix.
A common evidence foundation supports five workspaces: Patient, Support Circle, Clinician, Care & Costs, and Discovery Lab. Patients’ observations can inform the timeline, progress views, Twin and clinician packet. Families can also make travel, spending and caregiver time visible.
What the working MVP demonstrates
The public application contains 65 fictional patient histories and demonstrates longitudinal evidence review, trackers, Digital Twin exploration, public-context comparison and clinician packets. Context Discovery uses prepared email, photograph and caregiver examples to demonstrate how candidate observations are reviewed, accepted or dismissed. These examples establish the review workflow; permissioned live extraction is a subsequent integration step.
The engineering approach combines structured observation models, deterministic consistency controls, provenance and human review. AI and multimodal development focus on converting unstructured material into traceable candidate observations. Supporting documentation details architecture, data engineering, testing, safeguards and implementation status.
The next phase will evaluate these workflows with patients and clinicians, measuring information completeness, clinician review effort, patient understanding and follow-up completion, while advancing permissioned integrations.
WholeContext’s contribution is continuity with accountability: patients retain a voice in their history, clinicians can investigate the evidence behind it, and both can follow whether care is helping the person return to everyday life.
Working MVP: https://wholecontextai.kchulis.chatgpt.site/
Solution Overview (MVP Demonstration):
Data & AI Solution
Live demo, prototype, or working product
WholeContext AI™ — Interactive MVP Demonstration
Primary URL: https://wholecontextai.kchulis.chatgpt.site/
Explore the WholeContext AI platform, including its five connected workspaces, Context Discovery, WholeContext Digital Twin™, longitudinal health and treatment-response tracking, clinician evidence intelligence, and OpenDataDelta™ federal-data integration.
The demonstration features interactive patient journeys using fictional health records.
If the branded domain cannot host the application directly, https://wholecontext.ai will provide backup access
Access instructions (if needed)
If the branded domain cannot host the application directly, https://wholecontext.ai will provide backup access. Email kchulis@gmail.com for technical assistance viewing the site.
Results for the American People:
Americans facing Lyme disease, Long COVID and complex chronic conditions should not have to rebuild their health story each visit. WholeContext AI connects patients, caregivers and clinicians through a patient-reviewed history.
CDC estimates 476,000 Americans are diagnosed and treated for Lyme disease annually. Our six-person U.S. patient sample prioritized chronology, memory and information control, shaping correctable histories and approval before sharing.
Patients turn dated records, symptoms, pain, treatments and daily function into a source-linked, two-page clinician brief. Caregiver observations remain attributed. Clinicians investigate evidence and open questions; patients track changes alongside treatment timing to inform follow-up and adjustments.
The public MVP includes 65 fictional histories. In Marion’s journey, WholeContext Digital Twin™ connects her goal of dressing comfortably with symptoms, function and interventions. Her clinician can trace changes after treatment to dated observations.
Stored CDC, HRSA and Census-derived context adds county Lyme reporting, physician supply and transportation barriers to inform exposure questions and care planning. Population context is distinct from diagnosis. Families track spending, travel and caregiver time.
The benefit is continuity: lived experience becomes evidence patients can correct, clinicians can investigate and families can carry forward, so each encounter builds on the last.
Data, Tech & Evidence Driven Approach
WholeContext is not another AI medical summary. It is a longitudinal evidence system that connects scattered information, preserves its origin, and updates a shared patient story as new facts emerge.
Its observation spine brings together records, symptom diaries, treatment and function measures, caregiver input, and reviewed communications. Dates, sources, corrections and uncertainty travel with each observation into patient progress views, the Digital Twin and clinician packets.
We researched 171 data sources and candidate resources across disease surveillance, environment, care access, economics and clinical research. The MVP applies selected U.S. public-data snapshots, including CDC surveillance and HRSA care-access information. OpenDataDelta™ isolates their contribution: what changes when a relevant layer is added or removed, and what does not? Geography, time period and provenance remain visible; population context is never mistaken for an individual finding.
Our hybrid approach uses deterministic logic for chronology, reconciliation and source comparisons; bounded, review-first AI-assisted categorization; and evidence checks that distinguish support, contradiction and missing information. Reversible Digital Twin scenarios explore what additional evidence might change without rewriting the history.
Across 65 fictional cases, 22 automated suites passed checks covering identity, chronology, provenance, corrections, source comparisons, scenarios and exports.
The result is inspectable, correctable intelligence that supports clinical review, coordinated follow-through and tracking progress over time.
The architecture includes a FHIR/USCDI-aligned interoperability roadmap for sharing source-linked evidence across conditions and care settings.
Execution, Efficiency & Delivery
An AI answer is not a health outcome. Value comes when the right evidence reaches the right person, survives the next handoff, and helps track whether care is working.
In eight weeks, one founder delivered a five-workspace MVP built around one source-linked history. Patients correct their story; clinicians review dated evidence, gaps and competing explanations; care partners coordinate authorized support. The same observations power the Digital Twin, progress and treatment-response views, clinician packets and cost scenarios. OpenDataDelta shows what relevant federal context adds.
The engineering evidence is concrete: 22 automated suites passed across 65 fictional cases, checking provenance, corrections, historical cutoffs, permissions, scenario isolation and PDF consistency. Context Discovery reconciles reviewed email, photo and caregiver observations without duplicating measurements. Versioned checkpoints make changes testable and recoverable.
The core workflow is built, not merely proposed. Our 60–90-day target for focused real-user availability reuses this foundation while completing protected hosting, independent security and clinical review, and supervised evaluation.
Success is measurable: case preparation and review time, critical facts retained, completed handoffs, follow-up closure, patient-reported function and cost per usable case. Each release will be judged by value delivered, not features added.
Scalability, Reuse & Societal Impact:
WholeContext starts with Lyme: roughly 476,000 Americans are diagnosed and treated annually. Expansion addresses invisible illness: HHS estimates about 18 million adults have Long COVID; a U.S. study estimates over 15 million with diagnosed autoimmune disease. These populations overlap. The broader need is national: CDC estimates 194 million U.S. adults have a chronic condition; 130 million have multiple conditions. People with chronic and mental health conditions account for 90% of $5.3 trillion in annual U.S. health spending. WholeContext reuses one longitudinal evidence spine for patient history, clinician review, treatment-response tracking, caregiver support and costs. Condition-specific methods and geographic data adapt the platform rather than requiring a new application. Patient-led visit preparation and Discovery Lab create entry points; clinics, employers and health plans provide planned distribution. A 60-90-day focused-release target advances through production and pilot gates, with FHIR/USCDI exchange on the roadmap. At 1% reach among adults with multiple conditions, 1.3 million people could benefit. If pilots establish $100 in net annual value per user, that scale represents $130 million annually. We will measure review effort, follow-through, function and total cost. International expansion follows localized evidence and governance: WHO estimates 1.71 billion people live with musculoskeletal conditions alone.
Kimberly, I appreciate two choices in WholeContext: the patient can correct their own history while the source of each observation stays attached, and hypothetical Digital Twin scenarios stay separate from the record. It’s also good to see another solo build in this sprint that treats population data as context and never as a diagnosis.
Your Care & Costs workspace, where families make spending, travel and caregiver time visible, has a published reference point for the work-time part. Liu-Galvin et al. (J Am Board Fam Med 2025) used national survey data on full-time U.S. workers in 2022 and found a mean lost-labor cost of $1,944 with Long COVID, $1,466 with COVID but no Long COVID, and $702 for people who never had COVID (chart attached). It’s self-reported missed workdays times wages, so it covers the patient’s own time, not a caregiver’s, and Long COVID is one of your expansion conditions. Still, it shows a scale your cost scenarios could be checked against.
My entry turns published papers into simulators you can change and check against federal data, so a reader can see how each number moves: https://invisibleillness.crowdicity.com/post/896844
Isaac’s LymeHQ credits a family caregiver’s hours to the person who gave them, which resembles your attributed caregiver observations, and Jordan’s Cerewell also tracks treatment response in a longitudinal record. Isaac, Jordan, I’d be glad to hear whether you see overlaps with Kimberly’s approach.
Wishing you a successful pilot.
Users tagged:
Reply 0
Kimberly, I really appreciate how much control you give the patient over their own history. Being able to review and correct it, keep caregiver observations attributed, and still preserve where each observation came from, that feels like respect built into the design, not added on later. The Digital Twin part caught my eye too, especially that historical views preserve what was known at an earlier point and hypothetical scenarios stay separate from the record. My question is about corrections, when a patient corrects an entry, does the original stay visible alongside the correction, or does the corrected version replace it? I ask because I have seen how much trust depends on being able to trace what changed and why.
Reply 0