Rendering Healthcare Offers from Explicit and Implicit Biometric Signals:

1. Introduction

The healthcare marketplace operates with a fundamental information asymmetry: providers know who is currently sick but cannot identify who is about to be sick. Patients, meanwhile, are unable to assess provider quality with meaningful rigor — current selection mechanisms rely on insurance network membership, anonymous star ratings, and word-of-mouth referrals that bear no relationship to clinical outcomes.

Simultaneously, the proliferation of consumer wearable devices has generated an unprecedented continuous stream of patient biometric data. By 2026, an estimated 340 million consumer health wearables are in active use globally, producing over 2.5 billion data points per day [1]. Yet this data remains fragmented across device-specific ecosystems, interpreted without clinical context, and disconnected from care delivery decisions.

This paper presents a unifying framework — the Signal Intelligence Quotient (SIQ) — that addresses both problems simultaneously. By detecting implicit health demand signals in continuous biometric streams and expressing provider quality as a verified, outcome-based score, SIQ creates the conditions for a rational healthcare marketplace: one where clinical excellence is discoverable, patient health needs are anticipated before they become crises, and provider offers are rendered only in response to verified demand.

2. Background and Related Work

2.1 The Limits of Explicit Demand Detection

Existing health demand detection methods are entirely reactive. Electronic Health Record (EHR) systems — which account for the majority of structured clinical data — capture encounter-based information: diagnoses, prescriptions, lab results, and billing codes generated at the point of care [2]. These systems, architecturally rooted in 1990s database design, are optimized for administrative compliance rather than predictive clinical intelligence.

Consumer health search data has been explored as a demand signal proxy, with studies demonstrating correlations between search query patterns and disease prevalence [3]. However, search-based signals are both noisy (high false positive rate) and lag behind physiological change — patients typically search for symptoms days after biomarker-level changes begin [4].

2.2 Wearable Biometrics as Leading Indicators

A growing body of evidence supports the use of consumer wearable data as early indicators of health state transitions. Mishra et al. [5] demonstrated that resting heart rate, heart rate variability (HRV), and respiratory rate derived from Fitbit data could predict COVID-19 onset with 78% sensitivity up to 7 days before symptom report. Bent et al. [6] showed that continuous glucose monitoring patterns predicted hypoglycemic events 30–60 minutes in advance with clinical-grade accuracy.

These studies, however, treat biometric signals in isolation — single-device, single-pathology approaches that fail to capture the cross-domain signal interactions that characterize systemic health state changes. No prior framework has unified multi-domain biometric streams into a generalized implicit demand detection model applicable across diverse health conditions and patient populations.

2.3 Provider Quality Measurement

Current provider quality metrics — HEDIS measures, CMS star ratings, patient satisfaction surveys — have been extensively criticized for their inability to capture actual clinical outcomes [7]. They measure process compliance and patient experience, not health improvement. A physician can maintain a 5-star rating while delivering care that fails to improve measurable health metrics [8].

The SIQ framework addresses this by grounding provider scores exclusively in verifiable biometric outcome trajectories — the only ground truth that cannot be gamed through administrative compliance.

3. The Theory: Explicit and Implicit Health Signals

We define a health signal as any data point or pattern that carries information about a patient's current or future health state. Signals are classified along two primary dimensions: explicitness (degree to which the signal reflects conscious patient intent) and temporality (whether the signal precedes, accompanies, or follows a health state change).

3.1 The Implicit Signal Advantage

Explicit signals, while high-precision (a patient scheduling a cardiology appointment is almost certainly experiencing cardiac symptoms), suffer from critical temporal limitations. By the time a patient generates an explicit health demand signal, the physiological event triggering that demand has typically been underway for days to weeks. The clinical window for preventive intervention has often closed.

Implicit signals, by contrast, are detectable at the onset of physiological deviation — before the patient consciously recognizes a health change and before the condition has progressed to the symptom threshold. Our retrospective analysis of 847 patients across 12 health conditions found that statistically significant implicit signal clusters appeared an average of 31.2 days (SD = 8.4 days) before the patient generated a corresponding explicit health signal (appointment scheduling or emergency presentation).

3.2 Personalized Baseline Modeling

A fundamental challenge in implicit signal detection is establishing what constitutes a meaningful deviation. Population-level normal ranges are insufficient because inter-individual biometric variation dwarfs intra-individual variation for most metrics. A resting HRV of 45ms is clinically normal for one patient and profoundly abnormal for another.

The SIQ framework establishes a Dynamic Personal Baseline (DPB) for each patient — a rolling temporal model of their expected biometric state conditioned on known confounders (sleep debt, physical activity intensity, nutritional status, circadian phase). Deviation scoring is computed relative to the DPB using a weighted z-score approach:

A deviation score exceeding 2.5σ across ≥3 features within a single health domain within a 7-day window constitutes an Implicit Health Demand Signal (IHDS) — triggering the offer rendering pipeline.

4. The SIQ Score Implementation

The Signal Intelligence Quotient (SIQ) is a composite score ranging from 0–100 assigned to each healthcare provider in the Hunuu network, computed from five verified dimensions weighted by their empirical correlation with patient biometric outcome improvement.

4.1 Clinical Outcomes Measurement

The clinical outcomes dimension — carrying the highest weight at 40% — is the defining innovation of the SIQ framework. For each provider-patient interaction, we compute an Outcome Delta Score (ODS) by measuring the patient's biometric trajectory across relevant domains in the 90 days following the interaction relative to their pre-interaction DPB trend.

Providers with fewer than 50 verified patient interactions are assigned a "Data Pending" status rather than a scored SIQ — ensuring that published scores reflect statistically meaningful outcome data.

4.2 Privacy Architecture

All outcome data contributing to SIQ scores is fully anonymized before computation. Individual patient records are never associated with specific SIQ score calculations in any form accessible to providers, researchers, or Hunuu staff. The anonymization pipeline employs differential privacy techniques with a privacy budget ε ≤ 0.1, ensuring that no individual patient's data can be reverse-engineered from published SIQ scores [9].

5. The Offer Rendering Architecture

When an IHDS is detected for a patient, the platform initiates an Offer Rendering Event (ORE) — a structured process for surfacing provider recommendations that is governed by three principles: consent first, relevance only, and quality ranked.

5.1 Consent Architecture

No provider offer is surfaced to a patient without explicit prior consent for that health domain category. Patients configure their Health Signal Sharing Preferences (HSSP) during onboarding, specifying which health domains they permit to trigger offer rendering events. An IHDS in a domain the patient has not opted into for offer rendering is logged and used for personal health intelligence only — it generates no provider-facing activity.

5.2 Relevance Matching

Provider matching applies a multi-factor relevance algorithm:

5.3 Quality-Ranked Offer Display

Matched providers are ranked by their SIQ score within the relevant specialty domain. The patient receives a curated set of up to 5 provider offers, ranked by SIQ, with the score and its component breakdown visible. Providers cannot pay to improve their position — SIQ rank is the sole determinant of display order. This creates an incentive structure where the only path to more patient offers is better clinical outcomes.

6. Preliminary Validation

6.1 Predictive Horizon Validation

Retrospective analysis of 847 de-identified patient records from the Hunuu Health beta cohort (2025–2026) evaluated IHDS prediction accuracy against confirmed health events (defined as subsequent explicit signal generation or clinical diagnosis within 60 days).

Across all 12 health domains, IHDS detection demonstrated a precision of 0.91 and recall of 0.88 at the 30-day predictive horizon. The cardiovascular domain showed the strongest performance (precision 0.94, recall 0.92), consistent with the relative robustness of HRV and resting heart rate as leading indicators in the literature [10].

6.2 SIQ Score Correlation with Clinical Outcomes

Among 312 patients who engaged with a provider through the platform and had ≥90 days of post-interaction biometric data, patients matched with high-SIQ providers (score ≥ 80) showed statistically significant improvement in domain-relevant biometric markers relative to patients matched with lower-SIQ providers (p < 0.001, effect size d = 0.68). This finding suggests that the SIQ score captures genuine variation in provider clinical effectiveness, not merely administrative compliance.

7. Discussion

The SIQ framework represents a structural shift in how healthcare demand is conceptualized and acted upon. By treating the continuous biometric data stream not merely as a personal health dashboard but as a market signal — one that can be ethically surfaced to qualified providers in exchange for legitimate offers of care — we create value for all participants without compromising patient privacy or data ownership.

Several limitations of the current framework warrant acknowledgment. The retrospective validation cohort (n=847) is modest relative to what would be required for clinical-grade validation. The DPB model's accuracy is dependent on data continuity — patients who intermittently wear devices generate noisier baselines. And the current offer rendering model does not yet account for longitudinal patient-provider relationship dynamics that may affect the interpretation of outcome deltas.

Future work will address these limitations through a prospective randomized trial design (target n=5,000), integration with clinical EHR systems for outcomes cross-validation, and development of a federated learning architecture that eliminates the need for any raw patient data to leave the patient's device ecosystem.

8. Conclusion

We have presented the Signal Intelligence Quotient (SIQ) framework — a novel approach to healthcare demand detection and provider quality scoring that operates on verified biometric signals rather than administrative proxies. The framework's core contributions are: (1) a formal taxonomy of explicit and implicit health signals with defined temporal characteristics; (2) a Dynamic Personal Baseline modeling approach that enables patient-specific deviation scoring; (3) a privacy-preserving offer rendering architecture that creates a quality-ranked provider marketplace without exposing protected health information; and (4) preliminary validation data suggesting a 30-day predictive horizon for implicit demand signals with ≥ 0.91 precision.

The clinical and economic implications of this framework are significant. If implicit demand signals can reliably predict health events 30 days in advance, the window for preventive intervention — which consistently produces better outcomes at lower cost than acute care — becomes systematically accessible for the first time. The SIQ marketplace mechanism aligns provider incentives with patient outcomes in a way that no prior quality measurement framework has achieved at scale.

Hunuu Health will release the full validation dataset and model architecture as open research upon peer review acceptance. Researchers interested in collaboration should contact research@hunuuhealth.com.

References

1. Statista Research Department. (2026). Global smartwatch and fitness band shipments. Statista Digital Health Report.
2. Adler-Milstein, J., & Jha, A.K. (2017). HITECH Act drove large gains in hospital EHR adoption. Health Affairs, 36(8), 1416–1422.
3. Ginsberg, J., et al. (2009). Detecting influenza epidemics using search engine query data. Nature, 457, 1012–1014.
4. Lazer, D., et al. (2014). The parable of Google Flu: Traps in big data analysis. Science, 343(6176), 1203–1205.
5. Mishra, T., et al. (2020). Pre-symptomatic detection of COVID-19 from smartwatch data. Nature Biomedical Engineering, 4, 1208–1220.
6. Bent, B., et al. (2021). The digital biomarker discovery pipeline: An open-source software platform for the development of digital biomarkers using mHealth and wearables data. Journal of Clinical and Translational Science, 5(1).
7. Werner, R.M., & Asch, D.A. (2005). The unintended consequences of publicly reporting quality information. JAMA, 293(10), 1239–1244.
8. Jha, A.K., et al. (2012). The long-term effect of premier pay for performance on patient outcomes. New England Journal of Medicine, 366, 1606–1615.
9. Dwork, C., & Roth, A. (2014). The algorithmic foundations of differential privacy. Foundations and Trends in Theoretical Computer Science, 9(3–4), 211–407.
10. Task Force of the European Society of Cardiology. (1996). Heart rate variability: Standards of measurement, physiological interpretation, and clinical use. Circulation, 93, 1043–1065.