Digital Biomarkers Summit · Basel · 23 June 2026 · Poster

The audio processing changed. The reading held.

Our Basel poster, June 2026: under simulated changes to a phone’s noise cancellation, handcrafted speech measures shifted though patients had not; Cephalgo’s multi‑stream reading stayed stable and still distinguished patients from controls.

Request the poster Plan a stability audit

The poster

When the device changes mid‑trial: evaluating speech‑based digital endpoint robustness to on‑device preprocessing variability in longitudinal remote monitoring

Request the poster

Conference: 9th annual Digital Biomarkers in Clinical Trials Summit, hosted in partnership with Roche
Date: Tuesday 23 June 2026
Venue: Roche Building 1, Grenzacherstrasse 124, Basel, Switzerland
Data: A public Parkinson’s speech dataset, Parkinson’s disease versus healthy controls

What it shows

A silent change that can look like disease

The same recordings were processed several ways and each model was held fixed, so any movement in a reading came from the processing alone.

The audio changes first

A software update can change a phone’s built‑in noise cancellation. The audio changes, and speech measures computed from it can shift though the patient has not changed.

Easy to mistake for progression

In a long study the shift arrives without warning and can look like real disease change. A single type of speech measure cannot tell the two apart.

Stable, and still distinguishes patients from controls

Cephalgo’s multi‑stream reading combines linguistic, acoustic and learned audio streams. It was the only approach tested that stayed stable across recording conditions and also distinguished patients from controls well.

Handcrafted measures moved with the processing

Handcrafted acoustic measures still distinguished patients from controls, but their output followed the processing change.

What it means for your trial

Check the measures you rely on

If your trial records speech over months, a phone or software change can move your endpoint while the patient stays the same. The check runs on recordings you already hold, beside your current endpoint.

A stability audit

Where phone or software changes can enter your study, and whether your current speech measures move with them. Stability audit ›

Audio from a trial you already ran

A re‑analysis pilot reads recordings you already hold, on your trial’s own scale. Re‑analysis pilot ›

Your device plan

Separate from this poster: how Cephalgo keeps readings on one scale when phones or their software change, with study phones or BYOD. Device changes ›

Parkinson’s programs

What Cephalgo reads in de novo and early Parkinson’s trials. Parkinson’s ›

Our 2026 posters

From Copenhagen to Boston

Five posters in 2026, from explainable screening in Copenhagen to phone and software changes in Basel, Philadelphia and Boston.

Posters and white papers