September 23, 2026
How Long-Term Health Data May Reveal Hidden Health Correlations
Longitudinal health records may help users identify meaningful patterns over time. For users who have been collecting data for months or even years, it might be possible to look beyond the daily scores and recommendations based on heart rate, sleep, and steps to identify long-term trends. Users can explore potential correlations between medications, symptoms, vital signs, lifestyle factors, and other health data—on their own, with healthcare professionals, or even with the assistance of AI.
The value of longitudinal analysis is well established in healthcare research. For example, “Longitudinal analysis, which involves tracking changes in data over time, is a critical tool for understanding disease progression, predicting outcomes, and personalizing treatment strategies”.{1} MedM users have also shared real-world examples of how longitudinal health records have become valuable over time, sometimes in unexpected ways. We are delighted to share two of these case studies.
Case Study: Discovering Hidden Health Correlations Through Long-Term Data Tracking
In 2021, a MedM user turned to the platform to help manage his mother’s health, recording medications, blood pressure, blood glucose, and other measurements. He later started using the same app to track his own medications, supplements, and health events.
Over the following four years, he experienced recurring episodes of severe depression. Despite consultations with healthcare professionals, no clear explanation emerged. Because he had consistently logged his medication intake alongside significant health events, he was eventually able to review several years of historical data and identify a recurring pattern: each severe depressive episode followed a period during which he had been taking a beta blocker prescribed to manage stress-related rapid heart rate while caring for his mother.
The delayed onset of symptoms had obscured the potential relationship, making the pattern difficult to recognize without reviewing long-term records. After discussing the observation with his healthcare providers and discontinuing the medication under medical supervision, he reported that the severe depressive episodes did not recur: “Only by reviewing nearly four years of medication and health logs did I recognize the pattern. Without keeping those records, I don't believe I would ever have made the connection.”
This experience illustrates an important principle of personal health informatics: meaningful health correlations may emerge only when different types of health information are viewed together over extended periods. Whether explored manually or with the assistance of AI, comprehensive longitudinal health records can help patients and clinicians generate new questions, identify patterns worth investigating, and support more informed healthcare discussions.
Case Study: Providing Better Context at the Point of Care
A caregiver used MedM to maintain comprehensive health records for an elderly parent, including blood pressure, blood glucose, body temperature, medications, and other key health measurements. Before each appointment, the caregiver provided clinicians with printed graphs showing trends over the previous three months.
Rather than relying solely on measurements taken during a clinic visit, healthcare professionals were able to review changes and trends over time, providing additional context for evaluating the patient’s health and ongoing care: “One doctor told me he wished all his patients and caregivers could provide such meaningful records at every appointment.”
According to the MedM user, clinicians consistently welcomed the reports, commenting that having clear, longitudinal health records made it easier to put individual measurements into context and discuss the patient's progress during appointments.
Can AI Help Us Learn Something from Years of Data?
When provided with comprehensive longitudinal data, AI may help identify potential patterns, generate hypotheses, or highlight trajectories that might otherwise be difficult to notice across years of records. This potential is being explored across a growing body of healthcare research. In one study, AI is described as a tool that “significantly boosts diagnostics, treatment planning, and personalized care.”{2} Personal health records (with personally identifiable information removed) can also be explored using personal AI assistants such as ChatGPT or Claude. It is important to note that any observations generated in this way should be treated as starting points for further investigation and interpreted in consultation with qualified healthcare professionals.
AI is not guaranteed to find meaningful correlations. Sometimes an analysis may reveal no clear relationship in the available data, and this can also be informative. The goal of AI-assisted health analysis is not to generate answers at any cost, but to help users explore their health records from different perspectives, ask new questions, and identify observations that may deserve further attention.
You may not know which measurement, symptom, medication, or life event will become important in the future. Maintaining a comprehensive personal health record can provide valuable insight into health today while creating a foundation for users, healthcare providers, and future AI tools to answer questions that have not yet arisen.
Analyzing Your MedM Health Records with AI
MedM allows users to export their health records for further analysis. For those who have been collecting health data for a while, AI may help discover trends, correlations, or questions worth exploring further. The following step-by-step guide explains how to prepare and explore MedM Health records using an AI assistant.
Step 1. Export Health Data
Open the MedM Health Diary app our blood pressure monitoring app, or our blood glucose monitoring app, and navigate to: Export → All Data. Save the exported ZIP archive to your computer or mobile device.
Step 2. Review Exported Data
The ZIP archive contains individual CSV files for each health measurement type. For example: Blood Pressure.csv, Blood Glucose.csv, Weight.csv, Activity.csv, Medication.csv, Symptoms.csv. Even extensive health histories can remain surprisingly compact. For example, one seven-year-long record—including activity measurements imported from Garmin and Apple wearables—amounted to less than 500 KB of data.
Step 3. Protecting Privacy
Before uploading any files to an AI service, review them carefully and remove personally identifiable information that is not needed for the analysis, such as: name, Email address, phone number, physical address, other personal identifiers.
Keep in mind that health data is sensitive even when obvious personal identifiers have been removed. Review the privacy and data-handling practices of any AI service before sharing your health records.
Step 4. Upload the Files to Your Preferred AI Assistant
Open your preferred AI assistant (such as ChatGPT or Claude), upload one or more CSV files, and describe what you would like to explore.
An Example Prompt for AI to Analyze Biometric Data
I have attached several CSV files exported from a personal health record. They contain longitudinal health data collected over several years, including vital signs, medications, symptoms, activity, sleep, and other measurements. Please analyze the data for long-term trends, recurring patterns, and potential correlations between different measurement types. Pay particular attention to changes that consistently occur before or after symptoms, medication changes, or lifestyle changes. Identify observations that may warrant further investigation, explain why they stand out, and suggest additional questions or analyses that could help me better understand my health. Please distinguish clearly between observations supported by the data and hypotheses that would require further confirmation. If the data does not support a meaningful pattern or correlation, say so rather than proposing one. Do not make medical diagnoses.
AI assistants can be particularly useful when given a clear analytical framework. For example:
Analyze this data from the perspective of a health researcher rather than a physician.
Look for: recurring temporal patterns, measurements that rise or fall together, delayed correlations (days or weeks later), seasonal effects, medication or supplement periods associated with changes in symptoms, gradual trends over months or years, unusual outliers, measurements that consistently change before another measurement changes.
For every observation: explain why it may be meaningful, state how confident you are in the observation, suggest additional data or analyses that could strengthen or challenge the observation, and clearly distinguish correlation from causation.
Bridging the Gap Between Longitudinal Home Health Data and Healthcare
Research into shared longitudinal health records{3} has highlighted the value of consolidating information from disparate sources into a single, person-centric record, providing a more comprehensive longitudinal view of an individual’s medical history. The same principle becomes particularly relevant as more health data is generated outside the clinic.
Connected medical devices, wearables, and health apps are making it increasingly easy for people to collect health data in their everyday lives. On top of measurements taken during healthcare visits, home-collected data can provide a longitudinal view of what happens between appointments. Yet integrating data from wearables and other connected devices into existing healthcare infrastructures and making it useful to healthcare providers remains a challenge{4}.
For more than a decade, MedM has been developing apps and connected health solutions for people tracking health data at home. Today, the MedM ecosystem supports more than 30 types of health and wellness data and 1000+ connected Bluetooth devices. MedM apps have earned thousands of five-star reviews, with users frequently highlighting the ease of use. Data from connected devices, manual entries, and other health ecosystems can be consolidated into a longitudinal personal health record, giving users a more comprehensive view of their health data over time.
Bridging the gap between home monitoring and healthcare does not require every user to participate in a formal remote patient monitoring program. Individuals using MedM personal health and wellness monitoring apps can review their records over time and generate reports to share with their healthcare providers, bringing measurements collected between appointments into the conversation. Where remote monitoring is part of the care model, MedM's SaaS RPM platform can also make patient-generated health data available directly to authorized healthcare professionals.
As AI becomes increasingly capable of analyzing complex health data, the value of comprehensive, high-quality longitudinal records may grow further. Advancing predictive healthcare will depend not only on increasingly capable algorithms{5}, but also on making meaningful health data available in ways that can be used in real-world healthcare. Whether reviewed by users themselves, shared with healthcare professionals, or explored with AI, longitudinal health data collected at home can provide context that individual measurements cannot. The data collected today may help answer health questions that have not even arisen yet.
References:
- “AI for Longitudinal Analysis: Tracking Disease Progression Over Time” by Oluwaseyi Kolawole Oladele, March 2025.
- “Artificial Intelligence for Clinical Prediction: Exploring Key Domains and Essential Functions” by Mohamed Khalifa and Mona Albadawy, March 2024.
- “Shared Longitudinal Health Records for Clinical and Population Health” by David Broyles, Ryan Crichton, Bob Jolliffe, Johan Ivar Sæbø, Brian E Dixon, February 2016.
- “Integrating AI-driven wearable devices and biometric data into stroke risk assessment: A review of opportunities and challenges” by David B. Olawade, Nicholas Aderinto, Aanuoluwapo Clement David-Olawade, Eghosasere Egbon, Temitope Adereni, Mayowa Racheal Popoola, Ritika Tiwari, February 2025.
- "Artificial Intelligence in Predictive Healthcare: A Systematic Review" by Abeer Al-Nafjan, Amaal Aljuhani, Arwa Alshebel, Asma Alharbi and Atheer Alshehri, September 2025.
