{
  "id": 159614,
  "title": "PassiveDx: The Body's API",
  "url": "https://urgent.news/2026/08/05/passivedx-the-bodys-api",
  "topic": "health",
  "section": "Health & Medicine",
  "published": "2026-08-05T04:39:03.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/alirezaai/passivedx-the-bodys-api-4f3c"
  },
  "original_language": "en",
  "account": "PassiveDx is an innovative approach to healthcare that aims to detect deviations in a person's normal health state without requiring continuous monitoring or intervention. This system treats a person's behavioral and physiological baselines as an API, learning what \"normal\" looks like for each individual and identifying meaningful changes over time.\n\nThe core idea behind PassiveDx is that every person has unique patterns in their movement, gait, sleep rhythm, heart-rate variability, activity patterns, respiratory patterns, micro-movements, and interaction patterns. These patterns form a longitudinal signature that can be used to establish a personal baseline and detect anomalies when they occur.\n\nPassiveDx reverses the traditional approach to healthcare by focusing on detecting deviations rather than identifying specific diseases. Instead of asking \"Which disease does this person have?\", PassiveDx asks \"What has changed?\" This distinction is crucial in identifying health issues at an early stage, before symptoms become apparent.\n\nTraditional clinical systems are built around known diseases, following a path from symptoms to tests and diagnosis. In contrast, PassiveDx starts with continuous behavior analysis, establishing a personal baseline, and then detecting deviations from that baseline. This approach allows for the early detection of health issues, acting more as an early-warning system than an autonomous doctor.\n\nThe system relies on passive sensing, utilizing signals already collected by devices people use daily, such as smartphones. Smartphone keystroke dynamics, for example, have been studied as potential digital biomarkers for cognitive and neurological states. Research has also explored continuously collected typing metadata as a low-burden source of behavioral information.\n\nPassiveDx treats these signals as probabilistic evidence rather than diagnostic truth, recognizing that different populations and domains may exhibit varying relationships between these signals and cognitive outcomes. The system also considers Wi-Fi Channel State Information (CSI) as a potential sensing modality, enabling contactless sensing of respiratory motion using commodity hardware.\n\nThe architecture of PassiveDx consists of seven layers:\n\n1. Consent Fabric: Privacy is prioritized as the first layer, with users deciding which sensors are enabled, what data is processed locally, what metadata can leave the device, and when consent expires.\n\n2. Passive Sensors: PassiveDx can consume signals from devices like smartphones, including accelerometer data, gyroscope data, motion features, interaction patterns, keyboard typing speed and behavior, and wearables data such as heart rate, HRV, sleep patterns, and activity levels. Only data with explicit opt-in will be collected.\n\n3. Local Feature Engine: Raw data remains local whenever possible, with the system extracting features from the raw signal, detecting signal quality, performing context detection, applying a privacy filter, and creating a feature vector. Raw audio is discarded if acoustic features are not required.\n\n4. Personal Baseline: PassiveDx establishes a personalized baseline for each individual, rather than using static thresholds. The baseline adapts to changes in a person's life, such as travel affecting sleep patterns or exercise influencing heart rate.\n\n5. Multimodal Anomaly Fusion: The system employs anomaly detection algorithms that consider magnitude, persistence, modality count, signal quality, context, and baseline distance. This approach allows weak anomalies to be combined into clinically relevant anomalies when multiple independent signals move together.\n\n6. Federated Intelligence: The global model learns from populations, while the personal model remains adaptive and specific to each individual. This approach enables passive learning and continuous improvement of the system.",
  "summary": "Building an AI That Watches Without Asking You to Watch You don't check your health. Your health checks in with you. Most health-monitoring systems have the same fundamental assumption: The patient must participate. Open the app. Measure your heart rate. Take a blood pressure reading. Answer a questionnaire. Complete a cognitive test. Look at your dashboard. But human health does not behave like…",
  "key_points": [
    "PassiveDx detects health deviations without continuous monitoring",
    "Uses personal baseline to identify meaningful changes over time",
    "Relies on passive sensing from daily devices like smartphones"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}