Cognitive Twin for dementia at home

See at home what today is discovered in the emergency room

Acute decline in dementia is discovered late. We recognise it by comparing each person with their own non-stationary baseline, not with a population threshold.

Pre-seed round openClinical partnership with an Italian IRCCS

Request the investor deck How it works

The alerting functions described are in development and subject to clinical investigation and CE marking as Class IIa. Duomea is not CE-marked.

The problem

At home, an acute and reversible decline is mistaken for the disease progressing

Delirium, infections, dehydration, drug interactions: in a person with dementia they look like “a bad day” and are recognised late, often in the emergency room.

For the family

The caregiver sees changes every day but has no reference to tell normal from abnormal.

For the health system

Avoidable hospitalisations, inappropriate ER visits, early institutionalisation.

For the physician

Between visits there is no data: only the account, after the fact.

The thesis

The shape of change against the personal baseline separates acute from progression

Progression: slow, monotonic driftAcute episode: abrupt, fluctuating, reversible

Illustrative schema with synthetic data.

Two-timescale model

Slow component: the disease trajectory
Fast residual: the daily variation, with personal variance
Change-point on the residual for abrupt onset; slope of the slow component for drift
Signature classifier validated on adjudicated study events

The solution

A cognitive twin that lives in the patient’s home and keeps their data there

Voice interaction

A few minutes a day with an original voice-based cognitive test, in ever-new variants: it measures the person against themselves.

Discreet observation

Two ambient cameras processed locally: only activity and movement indicators, no recording.

Edge, not cloud

Voice, language and vision run on a unit in the home. Only encrypted measures leave. Privacy by design.

Clinical pathway

Goal for Class IIa: flag to the clinician, with evidence, a change compatible with an acute event. A person always decides.

Off-the-shelf tablet, two cameras and a local edge unit; configuration under evaluation. The architecture is open: any compatible device becoming a Duomea agent is a post-Class IIa roadmap item, not a claim about today’s product.

Where we are

Early, and we say so

The system is at TRL 3 today. The pre-seed takes it to TRL 4 and the edge unit to TRL 5; the patient study takes it to TRL 6.

Done

LOI signed with an Italian IRCCS
Software IP assigned from Deepleey to Duomea
Edge tests started on NVIDIA Jetson

Pre-seed goals

Written opinion on first-stage class, QMS started
Term sheet with the IRCCS, CEO search
Volunteers at home, test-retest reliability of the cognitive test

Seed goals

Home study with the IRCCS, adjudicated events
ISO 13485 certification
Class IIa pre-submission

Regulatory pathway

Two stages, because the claim defines the class and the class defines the value

Pre-seed

Qualification and QMS

Written opinion on first-stage class
ISO 13485 operational
No alerting claim

First stage

Structured collection

If qualified as Class I
Puts QMS, hardware and data into operation
Not the revenue thesis

Seed

Clinical study

Italian IRCCS, at home
Adjudicated events
TRL 6

Seed → A

Class IIa

Alerting the clinician, discriminating acute from progressive
Notified body
IIb risk checked in pre-submission

MDR Rule 11: software providing information for clinical decisions is at least Class IIa.

Clinical partner

The clinical study buys our hardest asset to copy: the data

LOI signed with an Italian IRCCS; term sheet under negotiation. Closing it is the first goal of the pre-seed.

How it is structured

Study sponsored by Duomea, paid on results, never on time
Home setting, same hardware as the product
Informed consent from the first patient, GDPR Art. 9

What Duomea gets

Ownership of the derived dataset and trained models
Consent covering commercial use from the first patient
Exclusivity on the dataset
No commercial veto, no uncapped royalties

Defensible advantage

A later entrant would have to redo the study: that is what makes the dataset an asset

What we own

Longitudinal home dataset, with consent for commercial use from the first patient
Trained models and foreground IP
Original cognitive item bank, in equivalent variants
Class IIa technical file and QMS

Against whom

Ambient monitoring for the elderly: falls and activity, no cognitive signature
Digital cognitive assessment: one-off clinic tests, no continuous home measure
Home care platforms: integrable, not competing on the claim
Consumer voice assistants: no clinical claim possible

Long-term vision

From monitoring to a companion for cognitive independence

Our long-term idea of intelligent home care for people living with dementia.

Sense

Tablet and ambient sensors learn each person’s own daily baseline.

Accompany

A mobile companion that follows the person and engages through the home.

Assist

A complete digital caretaker with a familiar voice and physical presence.

Vision concept, not a current product. Duomea is not CE-marked.

Business model

The institution pays; material revenue comes after the Class IIa pathway

Institutions first

The institution pays. A facility configuration for nursing homes lowers the cost per patient.

Families as a second channel

A direct offer for families who look after a person at home.

Italy, then Europe

Italy first, then France, Germany and Spain. Public reimbursement becomes possible after Class IIa.

The team

A team covering AI, edge, clinical and regulatory

Gabriele Torre

Founder & Chief AI

Models, ML pipelines, Cognitive Twin architecture.

Fabrizio Malfanti

CTO

Digital twin architecture, edge, technical and product leadership.

Clinical & Regulatory lead

Being finalised

MDR, clinical evaluation, PMCF. Owner of the study and the file.

Giuseppe Russo

Senior Data Scientist

Modelling of clinical and behavioural data.

Michela Paparella

Co-founder, Strategy & Ecosystem

Clinical and institutional ecosystem, caregivers, adoption.

Cristiano Ercolani

Commercial & Capital

Enterprise IT, fundraising, M&A transactions.

CEO

Search underway, joins at Seed

Regulated medtech, market access and reimbursement in Italy, Series A.

The round

Pre-seed of €500k, open

To close the clinical partner term sheet, hire the CEO, set up the QMS and take the system to TRL 4. Terms and materials are in the investor deck.

€500k

Pre-seed round

Investor information