Create Jira Ticket
This idea cannot be sent to Jira.
PathwayAI: Making the Hidden Patient Journey Visible
PathwayAI is developed by Yuwei Zhang, a data scientist trained in pathology and epidemiology. Her family’s experience with tick-related illness informs its focus on healthcare access and the hidden burdens patients and families face.
PathwayAI connects prevention, care preparation, support navigation, and economic burden insights. People can reuse their entries and download a patient-reviewed visit-and-cost summary, reducing the need to repeat their story. Community teams can explore public data and published research through a Hudson Valley pilot view comparing Dutchess and Orange counties in New York. Cost scenarios are illustrative, not measured county costs.
Its distinguishing focus is connecting personal experience with community planning. A future consent-based reporting pilot would build on existing tick surveillance and encounter-reporting tools by capturing the subsequent patient journey—including diagnostic delays, expenses, missed work, caregiver time, and care barriers. Privacy-reviewed summaries could help communities identify unmet needs and inform resource planning.
Solution Overview (MVP Demonstration):
User-Facing Application
Live demo, prototype, or working product
Working product—PathwayAI is a functioning MVP that people can use online. https://pathwayai.streamlit.app/
Access instructions (if needed)
Open https://pathwayai.streamlit.app/ or scan the QR code in the attached image.
To test Care & Support, use fictional information, organize your story, review the summary and download your visit-and-cost card. AI assistance for organizing your story is optional.
Select Community Burden & Action to explore county comparisons, cost illustrations and community resources.
Results for the American People
PathwayAI makes Lyme disease’s hidden burden visible—from diagnostic delays and expenses to missed work, disability and caregiver time. It helps people prepare for prevention and care, organize their story, find support and download a patient-reviewed visit-and-cost summary.
Complicated forms risk excluding people who are too unwell to complete them, including those needing assistance most. Optional AI helps organize their story for review, while reusable entries reduce repeated typing for people experiencing fatigue, dizziness or difficulty concentrating.
Policymakers and community teams can compare Dutchess and Orange counties using public data and published studies. The app clearly labels reported facts, research findings and illustrative cost estimates. Cost illustrations are not measured county costs.
Starting locally recognizes differences in healthcare access, language and available support. A future consent-based pilot would add patient experiences of diagnostic delays, healthcare use, expenses, missed work and caregiver burden. Privacy-reviewed summaries could inform community planning.
PathwayAI connects practical patient support with insight into burdens routine disease counts can miss, especially before diagnosis. Its goals are timely care, less financial and family burden, reduced avoidable disability and stronger community connections. These benefits require evaluation; savings and improved health outcomes have not been demonstrated.
Data, Tech & Evidence Driven Approach
PathwayAI combines federal open data, state surveillance, published research and patient-reported experience, with clear evidence labels.
Federal sources include CDC tick distribution and historical county Lyme case files, Census community data, HRSA Area Health Resources Files and CDC PLACES. These describe tick distribution, language, disability and healthcare resources. PLACES estimates are modeled. NYSDOH provides reported Lyme cases, rates and sampled tick observations; regional health assessments add local context. Tick sampling reflects collection sites, not countywide tick density.
The Hudson Valley cost illustration applies Hook et al.’s (2022) mean societal cost of $2,032 to 2024 reported cases—1,060 in Dutchess and 964 in Orange—yielding approximately $4.11 million combined in 2016 dollars. The study’s societal-cost analysis included 613 participants from four high-incidence states during 2014–2016. These scenarios illustrate scale, not measured county spending, current prices or costs specific to the pre-diagnosis period.
The Python/Streamlit app helps users organize their story, review entries and download summaries. Optional AI assists with permission. The Research & Strategy Agent retrieves curated public evidence; cost calculations use explicit arithmetic. Private patient summaries remain separate from public evidence analysis.
Published Biobank findings add follow-up context. Voluntary survey findings remain separate from county scenarios and are not representative of county populations. Downloads retain sources and assumptions. Observed findings, literature evidence and synthetic or modeled examples are clearly labeled. Local patient reporting remains a future consent-based pilot.
Execution, Efficiency & Delivery
PathwayAI brings scattered information together in practical tools: a prevention plan, a visit-and-cost summary, a personal cost-and-time visual, support resources and county comparisons.
Fatigue, dizziness and difficulty remembering details can make preparing for care harder. PathwayAI helps people organize their story, costs and questions into a record they can review, download and bring to providers. Reusing entries within the same session reduces repeated typing. Optional AI helps organize information with permission.
Community teams can explore reported Lyme rates alongside clearly labeled cost illustrations covering medical care, other expenses and lost productivity. Downloadable calculations and briefs support planning around prevention, care access and community needs. These illustrations are not measured county spending.
Patient and community feedback informed simpler language, navigation and reading options. Targeted code checks covered story-field extraction and preservation of user corrections. Full navigation and download workflows still require verification in the deployed app. These checks assess software function, not clinical effectiveness.
The MVP delivers personal summaries and evidence tools for community planning. A future consent-based pilot would evaluate usability, access to support and whether the information helps identify barriers and inform resource decisions. Health improvements and cost savings remain to be measured.
Scalability, Reuse & Societal Impact:
PathwayAI runs on Streamlit, with code maintained in GitHub. Its modules handle geography, evidence retrieval, testing education, story organization and cost calculations. More counties can reuse the workflow with compatible surveillance, local resources and reviewed assumptions. Other illnesses require disease-specific evidence.
The next step is a consent-based pilot with community organizations, clinics and public-health teams. People could contribute reviewed records to help communities understand care barriers, costs and support needs. Reusing entries would reduce repeated effort. Partners could guide resources and evaluate whether people reach support.
Personal entries remain in server-side session memory. Users can download records for later visits. External AI requires permission; shared collection is disabled by default. Community reporting requires secure storage, consent, access and retention rules, consistent periods and privacy-reviewed summaries.
Current cost illustrations combine county case counts with historical study costs. They do not yet account for regional prices or the full burden of persistent and undiagnosed illness.
Priorities include usability and accessibility testing, stronger local evidence and sustainable operations. The intended impact is timely care and less financial, work and family disruption. These benefits require evaluation.
Hello Yuwei, a small addition that makes your Hudson Valley cost illustration easier to compare with today’s prices: restated from 2016 to 2025 dollars with the BLS consumer price index (a factor of 1.34), the $2,032 per case becomes about $2,726, and the $4.11 million for Dutchess and Orange counties becomes about $5.52 million (chart attached). It’s general price inflation, not medical-care inflation, and it remains an illustration of scale and not measured county spending, as your page says clearly.
This kind of restatement is done by code in my entry, not by an AI model, for example $2,227 in 2022 dollars becomes about $2,450 in 2025 dollars in simulator 2 (Yu et al., JAMA Netw Open 2026), labeled as general inflation. My entry is here if you’d like to look: https://invisibleillness.crowdicity.com/post/896844
I’d also be glad to hear Samuel’s view, since his BurdenBridge Pennsylvania Lyme atlas takes the same care not to turn an external cost study into county-level totals, and Victoria’s, since her One Health Lyme Gap Atlas also works with county-level surveillance.
I like that your visit-and-cost summary stays under the patient’s control and that your cost scenarios are always labeled as illustrative. Best wishes for the Hudson Valley pilot.
Users tagged:
Reply 0