Inspiration

We were inspired by the growing impact of extreme heat, wildfire smoke, poor air quality, flooding, and severe weather on public health. Although these environmental hazards can affect entire communities, they pose an especially serious risk to older adults, people with chronic health conditions, individuals with limited mobility, and patients without access to safe housing, transportation, air conditioning, or clean indoor air.

At the same time, healthcare providers often manage thousands of patients and have limited time to determine who requires immediate assistance. General environmental alerts can warn people about dangerous conditions, but they do not identify which patients are most vulnerable, explain the specific barriers they face, or connect them with direct support.

We created Pidge to bridge the gap between environmental warnings and personalized healthcare. Our goal was to develop a system that could identify vulnerable patients before a climate-related event becomes a medical emergency and help healthcare teams respond more effectively.

What It Does

Pidge is a climate-health risk assessment and response platform that combines real-time environmental conditions with patient-specific information.

The system evaluates environmental factors such as local air quality, extreme temperatures, wildfire smoke, flooding, and severe weather. It then considers individual factors, including medical conditions, age, mobility, housing conditions, transportation access, communication needs, and accessibility requirements.

Using this information, Pidge calculates a personalized risk level and recommends an appropriate response. A lower-risk patient may receive a simple notification with preventive guidance. A patient at moderate risk may receive an AI-assisted check-in asking whether they have access to medication, air conditioning, clean indoor air, transportation, or another safe location. A high-risk patient, or someone who reports an urgent need, can be escalated for direct outreach from a community health worker or clinician.

Pidge also includes a provider dashboard where healthcare workers can view scheduled patients, identify high-risk individuals, understand why someone was flagged, monitor patient check-ins, and prioritize unresolved cases. This creates a clear path from environmental warning to personalized action.

How We Built It

We designed Pidge around three connected components: a risk-scoring engine, a patient check-in experience, and a provider dashboard.

The risk-scoring engine combines environmental conditions with patient health, housing, mobility, and accessibility information. It evaluates the severity of each factor and produces an overall risk level. We focused on making the results explainable so healthcare workers could understand why a patient was flagged instead of receiving a score without context.

The patient-facing check-in was designed to be simple and accessible. It gives patients clear information about the environmental hazard, explains why they may be at risk, and recommends practical actions. It also asks focused questions about immediate needs, such as access to medication, cooling, clean air, transportation, and safe housing.

The provider dashboard organizes patients according to urgency. It displays each patient’s risk level, contributing risk factors, check-in status, and recommended response. This helps clinicians and community health workers quickly identify patients who may need direct outreach.

We also developed a visual system that communicates urgency using labels, icons, explanations, and color. We avoided relying on color alone so that the interface would remain understandable for people with different visual and accessibility needs.

Challenges We Ran Into

Our biggest challenge was translating complex environmental and health information into recommendations that were accurate, understandable, and actionable. Environmental conditions can change quickly, while a patient’s vulnerability depends on several interconnected factors. We had to determine how to combine these factors without oversimplifying the patient’s situation.

Another challenge was balancing detail with clarity. Healthcare providers need enough information to make informed decisions, but too much information can slow down their response. Patients also need guidance that is specific enough to be useful without feeling confusing or overwhelming.

We addressed this by organizing information according to urgency and displaying the most important action first. Providers can still access additional details when they need to understand the reasoning behind a patient’s risk level.

We also considered the limitations of using AI in healthcare. Pidge is designed to support healthcare workers, not replace them. When a situation is urgent, uncertain, or outside the system’s capabilities, the patient is escalated to a qualified healthcare professional.

Privacy and security were also major considerations because Pidge uses sensitive health, housing, accessibility, and location information. A real-world version of the platform would require informed consent, secure authentication, encrypted data storage, controlled access, and compliance with healthcare privacy requirements.

Accomplishments That We’re Proud Of

We are proud that Pidge turns complex environmental and patient data into a clear and useful experience for both patients and healthcare providers.

Instead of treating every person in an affected area the same, Pidge recognizes that environmental risk depends on a person’s health, surroundings, mobility, and access to resources. This allows the platform to provide more personalized and relevant support.

We also created a connected response system rather than a basic alert application. Environmental data identifies a possible hazard, the risk-scoring engine determines who may be vulnerable, the patient check-in identifies immediate barriers, and the provider dashboard helps coordinate the appropriate response.

Pidge can also reduce unnecessary notifications by matching each response to the patient’s level of risk. This helps prevent alert fatigue while allowing clinicians and community health workers to focus their limited time on patients who require direct assistance.

Most importantly, Pidge is designed around action. It does not stop after warning a patient that conditions may be dangerous. It identifies what the patient may need and connects that need to a practical next step.

What We Learned

We learned that effective health alerts must be personalized, actionable, accessible, and connected to real support. General warnings may tell people that a hazard exists, but they do not always help individuals understand what they should do next.

We also learned that health vulnerability cannot be measured using medical information alone. Housing quality, transportation access, mobility, language, communication preferences, and access to resources can determine whether someone can follow health guidance. A recommendation is only useful if it accounts for the barriers a patient may face.

Another important lesson was the value of explainability. Healthcare workers need to understand why a patient has been identified as high risk before deciding how to respond. Showing the factors behind the risk score makes the system more transparent and supports informed human decision-making.

We also learned that AI should support communication and prioritization rather than replace professional medical judgment. Automated check-ins can help collect information and identify urgent needs, but clinicians and community health workers must remain involved in important healthcare decisions.

If we approached the project again, we would involve patients, clinicians, and community health workers earlier in the design process. We would conduct interviews and usability tests before finalizing the risk categories and check-in questions. Their feedback would help us determine which information is most useful, which questions may feel intrusive, and how Pidge could fit into existing healthcare workflows.

What’s Next for Pidge

Our next step is to partner with local healthcare systems, community health organizations, and public health agencies to test Pidge in real-world settings. We want to gather feedback from patients, clinicians, and community health workers to improve the platform’s accessibility, recommendations, and workflow.

We also plan to strengthen the risk-scoring model by integrating additional real-time datasets, including hyperlocal air-quality readings, heat-index forecasts, wildfire-smoke movement, flood warnings, severe-weather alerts, and power-outage information. Historical environmental and health data could also help us evaluate and improve the model’s accuracy.

Future versions of Pidge could support multilingual communication and several notification methods, including text messages, automated phone calls, and accessible web-based check-ins. These options would allow the platform to reach patients with different language, technology, and accessibility needs.

We would also like to expand public enrollment so individuals can voluntarily use Pidge even if they are not connected to a participating healthcare provider. Partnerships with local organizations could connect users with cooling centers, clean-air shelters, transportation services, medication assistance, emergency housing, and other community resources.

Before deployment, we plan to work with healthcare, legal, and third-party cybersecurity specialists to strengthen privacy and data protection. This would include informed consent, data minimization, encryption, role-based access, secure authentication, and clear policies explaining how sensitive patient information is collected and used.

Ultimately, our goal is for Pidge to become a trusted connection between environmental data, healthcare providers, community resources, and the people who are most vulnerable to climate-related health risks.