by Christer Nilsson, CEO, Replior
The Nocturnal Paradox and Statistical Noise
The patient’s voice is critical, but it doesn’t tell the whole story. Regulators require us to measure patient feeling, yet human recall is naturally biased. This is especially true in dermatology trials measuring scratch.
Asking a patient, “How much did you itch last night?” doesn’t make sense. Relying on memory for events that happen while asleep introduces blind spots, forcing us to measure efficacy with incomplete data. We call this the “Nocturnal Paradox.”
This noise has a real cost. In dermatology studies, the placebo is often a moisturizing vehicle that actually can improve the skin lesions, artificially inflating the placebo effect. Statisticians usually combat this noise by increasing the sample size. But adding patients is expensive and inefficient.
While we have clinical observations and patient perceptions, we are missing a crucial piece: the objective measurement of actual behavior.
The Solution: A Dedicated Scratch-Detecting Ring
To close this gap, we built a wearable sensor. Scratching is a complex, biomechanical finger movement, so a wrist-worn device won’t capture high-quality data. The sensor needs to be where the action is.
We developed a smart ring explicitly designed for clinical trials. Since people scratch with both hands, the system supports up to two sensors per hand. The ring is lightweight, waterproof, and uses swappable sizing braces, saving clinical sites from replacing the core hardware for every patient.
Daily charging kills patient compliance. To fix this, we gave the ring enough battery life to capture data continuously, 24/7. It also tracks body temperature and movement to verify wear-compliance.
Edge Computing and AI in Action
The ring uses an edge neural network trained to detect scratching and ignore normal hand movements.
Every millisecond, the AI assesses incoming signals. If a movement hits the threshold for a scratch, the ring sends just that specific chunk of data via a mobile app to our servers. Computing on the “edge” prevents the continuous streaming of high-hertz raw data, which saves battery life and cuts cellular roaming costs.
Once on the server, a secondary neural network analyzes the event to extract vital metrics:
- Duration: How long the scratch lasted.
- Frequency: The speed of the fingers moving (in Hertz).
- Pressure: How hard the fingers pressed into the skin (in Newtons).
- Amplitude: How far the fingers moved side-to-side.
- Nail Involvement: Whether the scratch utilized the fingernails.
Validation and Looking Ahead
We are starting our first analytical validation study, testing the ring on 35-50 participants with various itch conditions. Over three nights, we will compare the ring’s nocturnal scratch and sleep data against infrared camera footage and PSG (polysomnography). Our long-term goal is to refine the AI to confidently monitor daytime scratching as well.
We offer comprehensive analytics for monitoring trends through a subject dashboard. Here, you can view granular data, like sleep onset, total sleep time, and scratch bouts per hour, or export the raw events directly. Ultimately, clinical scientists will determine which metrics serve as the best biomarkers for treatment efficacy.
By measuring actual patient behavior, we can lower data variance, run smaller studies, and deliver higher statistical confidence. We provide the objective data streams so the scientific community can unlock the next generation of clinical breakthroughs.
Watch the full recording of the session below, and reach out today for your demo.
