Is the health scan on my insurance app actually accurate?

The rise of smartphone-based health scans in insurance applications represents a major shift in underwriting technology. For a growing number of applicants, the first step in the journey is no longer a conversation with an agent but holding their phone and looking into the camera for a 60-second scan. This raises a reasonable question from consumers: "Is the health scan on my insurance app actually accurate?" But for chief underwriting officers, actuaries, and reinsurers, that same question has a different context. It's not just about individual accuracy but about data integrity, model validity, and the actuarial soundness of risk assessment based on these novel data streams.

"A 2023 clinical validation study on remote photoplethysmography (rPPG) software, the technology used in most camera-based health scans, found a mean absolute error of just 1.06 beats per minute when compared against hospital-grade ECG equipment. This level of precision for a core vital sign like pulse rate is a foundational data point for building trust in contactless assessments."

The core question: insurance health scan accuracy

At the heart of the insurance health scan accuracy debate is the technology known as remote photoplethysmography (rPPG). This technology uses a smartphone's camera and light to detect minute changes in the color of light reflected from the skin. These changes correspond to the pulsing of blood through the vessels, allowing the software to calculate several key vital signs.

The core output is a photoplethysmogram-a wave form that is optically similar to the one produced by a standard pulse oximeter. Advanced algorithms analyze this waveform to extract biomarkers, including:

• Heart Rate
• Heart Rate Variability (HRV)
• Respiration Rate
• Blood Pressure
• Oxygen Saturation (SpO2)

The validation of these outputs is where the question of accuracy is answered. In numerous studies, such as the one published in the Journal of the American Medical Association, rPPG has been benchmarked against "gold standard" medical devices. For heart rate and HRV, the comparison is typically an electrocardiogram (ECG). Research has consistently shown that under controlled conditions, rPPG can achieve a Mean Absolute Error (MAE) of 2-5 beats per minute, with some modern systems achieving even higher accuracy, as noted in the statistic above.

However, it is crucial for underwriters and actuaries to understand the factors that can influence data quality. These include:

• Lighting conditions: Consistent, bright lighting is essential for a clear signal.
• User movement: The applicant must remain still during the measurement.
• Camera quality: Newer smartphone cameras provide a better signal.
• Skin tone: Algorithms must be trained on diverse datasets to ensure accuracy across all skin tones, a major focus of ongoing research and development by companies like Circadify.

While heart rate is a mature and highly accurate rPPG measurement, blood pressure and SpO2 are still developing. For underwriting purposes, these are often used as indicative signals rather than diagnostic measurements, helping to triage cases for further review.

Data Source	Immediacy	Cost	Applicant Burden	Data Granularity
Traditional Paramedical Exam	Weeks	High ($150-$250)	High (Scheduling, Exam)	High (Fluids, Vitals)
Attending Physician Statement	Weeks to Months	Medium ($100+)	Low to Medium	Variable, Unstructured
Pharmacy / Rx History	Seconds	Low ($5-$20)	Very Low	High (Specific, Longitudinal)
Contactless Health Scan (rPPG)	< 2 Minutes	Very Low (<$5)	Very Low (Uses own phone)	High (Waveform, Vitals)

Industry applications and reinsurer audits

For insurers, the true test of insurance health scan accuracy is not just technical performance but its acceptance by reinsurance partners. A successful accelerated underwriting program relies on a reinsurer's confidence that the new data sources are robust and the risks are well-managed.

Biometric underwriting data quality: what reinsurers audit first

When a reinsurer audits a digital underwriting program, their primary concern is data quality and consistency. According to a 2023 analysis by McKinsey, digital underwriting requires robust audit and risk controls. Reinsurers will scrutinize the entire data pipeline, from the point of capture on the applicant's phone to its use in a pricing model. They will ask for:

• Validation studies: Third-party validation of the rPPG technology against medical-grade equipment.
• Data dictionaries: Clear definitions of every data point captured.
• Performance metrics: Error rates, signal quality scores, and completion rates for the scans.
• Fraud detection: How the system prevents spoofing or gaming of the measurement.

Ensuring models are actuarially sound

Swiss Re, in its 2022 principles for alternative data, emphasizes that any new data source must be "actuarially sound." This means a carrier must be able to demonstrate a clear, statistically significant correlation between the data captured by the health scan and mortality/morbidity risk. An accurate heart rate measurement is useful, but its variability (HRV) can be a more powerful predictor of cardiovascular health and all-cause mortality. Actuarial teams must prove their models using this new data are predictive and are not introducing unforeseen bias.

Current research and evidence

The evidence base for rPPG is expanding rapidly. Early research focused on proving basic feasibility, but current studies are more sophisticated. A 2023 study by researchers at the University of South Australia explored the impact of different light sources on rPPG accuracy, finding that algorithmic adjustments could compensate for non-ideal conditions. This type of research is critical for ensuring the technology is robust enough for at-home, unsupervised use. Academic and commercial entities are continually publishing new validation studies, often registering them on platforms like ClinicalTrials.gov to ensure transparency. This public body of evidence is what allows actuaries and underwriters to build the business case for adoption.

The future of contactless biometrics in underwriting

The future of insurance health scan accuracy lies in both software and hardware improvements. As smartphone cameras become more powerful, the raw signal quality will improve. Simultaneously, machine learning models are becoming more adept at filtering out noise from user movement or poor lighting. In the near future, we can expect to see an expansion of the biomarkers that can be reliably measured, potentially including early indicators for conditions like hypertension or sleep apnea. The end goal is not to replace the underwriter but to provide them with a richer, more immediate dataset to make faster, more confident decisions.

Frequently asked questions

Q: How accurate are phone camera health scans compared to a doctor's visit? A: For core vitals like heart rate, the best-in-class technology is very accurate, approaching the precision of clinical-grade devices like ECGs. However, they are intended for risk assessment and triage, not for medical diagnosis. They augment, rather than replace, traditional medical evaluation when high-risk factors are detected.

Q: What data do these insurance health scans actually capture? A: The scans capture raw video of the user's face, which is processed to create a photoplethysmographic (PPG) waveform. From this waveform, algorithms extract biomarkers such as heart rate, heart rate variability (HRV), and respiratory rate. Some may also provide indicative measurements for blood pressure and oxygen saturation.

Q: Can an applicant "fail" a health scan? A: An applicant doesn't "fail" in the traditional sense. A scan can result in a "no-read" or poor-quality signal if the conditions aren't right (e.g., too much movement or bad lighting), which might require the applicant to repeat the scan. If the scan successfully captures vitals that fall outside the insurer's thresholds for a given program, the application is typically routed to a human underwriter for further review, potentially requiring more traditional evidence like an APS or paramedical exam.

Q: How do reinsurers verify the insurance health scan accuracy? A: Reinsurers conduct rigorous audits of the primary carrier's underwriting program. They review the technology's validation studies, data dictionaries, and the predictive power of the biometric data against actual mortality and morbidity outcomes. They need to see a clear, auditable trail from data capture to risk assessment.

As accelerated underwriting becomes the industry standard, the reliance on technologies like remote photoplethysmography will only grow. For underwriting leaders and actuarial teams, the focus must be on rigorous validation, transparent data governance, and strong partnerships with reinsurers. Circadify is at the forefront of providing the tools and data infrastructure to address these challenges. To learn more about our actuarial data and validation methodologies, explore our resources for payers and insurers at circadify.com/industries/payers-insurance.