MIRA ANALYTICS

AI driven, real-time, precise and explainable data oversight for endpoints in clinical trials.

What we do

Automated scale ratings

From endpoint capture to AI-driven scoring.

Raters administer assessments and Mira ingests questionnaire audio locally and outputs item-level and total scale scores, enabling continuous oversight of rater performance.

Interview waveform flowing into a Mira engine and bar chart of scale ratings.

Explainability

Scores that point back to the interview.

Mira's scoring can be traced back to specific parts of the scale interview. Explainability helps clinical teams understand which parts of the dialogue drove the model's judgement — making reviews faster, more consistent, and easier to trust.

How have you been sleeping lately? 09:14

Not terrible, but not great either. 09:15

How long does it usually take you to fall asleep? 09:19

On good nights maybe half an hour. Sometimes longer, but nothing extreme. 09:21

And on the bad nights? 09:25

I’m awake until three or four almost every night, and when I do sleep I’m up again after an hour. 09:27

Mira Model output Short, fragmented sleep The participant reports frequent nights with delayed sleep onset and repeated awakenings — a pattern typical for severe insomnia.

How do these nights affect your day? 09:33

I can’t focus in meetings and I’m forgetting simple things. I feel wiped out most days. 09:35

Mira Model output Daytime cognitive impact The participant mentions reduced concentration and mistakes with routine tasks — evidence that their insomnia is functionally impairing.

How’s your mood and appetite otherwise? 09:40

My mood’s okay, and I still go for walks. Just wish I wasn’t so tired. 09:43

Mira highlights parts of the insomnia dialogue that support a severe rating.

Rater analytics

From single predictions to rater performance.

Individual scores aggregate into the intuitive analytics of rater behavior, and can highlight issues such as drift, over- or under-scoring tendencies. These metrics can be viewed on an individual assessment basis or longitudinally as they evolve throughout the study.

Monitor rater scores over time to visualize trends.

Platform

All the detail in one secure platform

A dashboard with insights for all your team's needs — built with security and compliance best practices in mind.

Security: Runs in a Virtual Private Cloud. Study data stay within dedicated secure infrastructure.

Flexibility: Can be utilized with any interviewer-rated questionnaires or scales. One engine that can travel across languages, sites and accents.

Insights: Understand the reliability and validity of key endpoints. Review scores, explanations, model confidence, and rater analytics — all in a single dashboard.

Tension MILD

Insomnia SEVERE

Anxiety MODERATE

Appetite ABSENT

Memory MILD

How have you been sleeping lately? 09:14

Not terrible, but not great either. 09:15

How long does it usually take you to fall asleep? 09:19

On good nights maybe half an hour. Sometimes longer, but nothing extreme. 09:21

And on the bad nights? 09:25

I’m awake until three or four almost every night, and when I do sleep I’m up again after an hour. 09:27

Mira Model output Short, fragmented sleep The participant reports frequent nights with delayed sleep onset and repeated awakenings — a pattern typical for severe insomnia.

How do these nights affect your day? 09:33

I can’t focus in meetings and I’m forgetting simple things. I feel wiped out most days. 09:35

Mira Model output Daytime cognitive impact The participant mentions reduced concentration and mistakes with routine tasks — evidence that their insomnia is functionally impairing.

How’s your mood and appetite otherwise? 09:40

My mood’s okay, and I still go for walks. Just wish I wasn’t so tired. 09:43

Individual item scores are estimated based only on the audio recording of the questionnaire.

Our results

The founders have a long track record of building internal AI tools for the pharma industry — including the oversight platform used at MindMed to help monitor key endpoints (HAM-A and MADRS) in their Phase 3 clinical programs.

The work “Using Large Language Models for Endpoint Oversight” received the Distinguished Poster Award at ISCTM 2025.

95.2% accuracy on central rater training.
1.57 (± 1.39) point average difference vs. central raters in Phase 2b.
Deployed in ongoing Phase 3 to monitor HAM-A & MADRS.

“Using Large Language Models for Endpoint Oversight”, ISCTM 2025 — poster · abstract

Scatter plot of Hammy vs central raters.

Founders

Adam

Co-Founder & Chief Executive Officer

Adam is a seasoned machine-learning specialist with over a decade of experience building ML products in both startup and corporate environments. He has worked in leadership roles, bringing machine-learning solutions, scalable data architectures, and practical AI applications to real-world problems.

in LinkedIn

Miguel

Co-Founder & Chief Technology Officer

Miguel holds a Ph.D. in Artificial Intelligence, having also over a decade of experience working across multiple healthcare startups as a machine learning engineer and technical lead, building AI-driven products and scalable data and analytics systems for clinical and healthcare applications.