Dormi

Dormi, Artificial Intelligence for sleep analysis

Our AI,
within API reach

Developed by Sleepacta after years of research in academia, Dormi is an algorithm built on a neural network trained and tested on a dataset of more than 6,500 real-life actigraphic examinations that required more than 17,000 hours of work by our data scientists.

Used daily in hundreds of examinations and then improved daily, the algorithm is capable of performing accurate screening of the sleep-wake cycle (including disorders and pathologies, such as insomnia and respiratory disorders during sleep) to support diagnosis in a non-invasive and inexpensive way.

Running on a certified, highly scalable cloud infrastructure to ensure optimal performance regardless of workloads, Dormi analyzes raw data recorded by simple and non-bulky wearable activity trackers commonly used for monitoring physical activity. This data, once processed by the Dormi algorithm, is queryable via authenticated REST APIs that can be easily integrated into any existing platform system. No personal patient data is managed or stored on the cloud infrastructure, thus ensuring full GDPR compliance.

Panoramica

Dormi, our neural network, as-a-service

Dormi is an algorithm that is the result of 10 years of research, involving doctors, bioengineers and artificial intelligence experts

Dormi has been tested on low-cost “commercial” wearable activity trackers (Fitbit and others) and validated against the gold standard (polysomnography) in over 500 recordings

Dormi is available via REST API

Dormi is based on an optimized neural network architecture, trained on more than 6,500 actigraphic examinations

Dormi is a CE Class I certified medical device

Dormi analyses and transmits the collected data in accordance with the GDPR for the protection of sensitive patient data

Actigraphy at the heart of the Dormi project

The actigraphs from which the Dormi algorithm acquires data are small smart devices designed to collect the wearer’s physiological parameters. Similar to a wristband or a wristwatch, and thus easily worn for long periods of time, they contain extremely sensitive three-dimensional motion sensors that enable them to distinguish and record periods of presence or absence of movement.

Thanks to an actigraph, it is therefore possible to measure the quality and quantity of sleep within a 24-hour circadian cycle and over several days, distinguishing the total hours of sleep, the duration of wakefulness, awakenings during the night and any “naps” taken during the daytime hours.

This makes these devices particularly suitable in the study and diagnosis of sleep alterations, whether related to or dependent on particular conditions (subjects subjected to shift work or jet lag) or disorders and pathologies (insomnia, hypersomnia, infantile hyperreactivity, Parkinson’s tremors).

In addition to identifying the phases of the sleep and wake cycle, some types of actigraphy can record a whole range of additional data, such as peripheral body temperature, surrounding noise and light levels, heart rate and variability, and the level of blood oxygenation in peripheral vessels, which enable even more accurate monitoring.

A certified, reliable and accurate algorithm

Dormi demonstrated more than 90 per cent concordance in sleep-wake rhythm classification compared to polysomnography, the gold standard in diagnosing sleep disorders, a method, however, that has huge limitations as it is complex, expensive, and invasive.

In addition, the continuous collection of data processed by Dormi helps to feed the neural network architecture and to constantly refine the accuracy of the algorithm. Being based on signal analysis tools using artificial intelligence and machine learning principles, the algorithm is therefore able to learn, process and classify data with a high degree of accuracy compared to polysomnography.

Certification

Validation and clinical certification

The Dormi algorithm applied to accelerometric raw data has been validated by many clinical studies and has been compared to polysomnography (to date the gold standard), both in healthy subjects and those with known sleep disorders. Thanks to this, since 2018 Dormi is a CE-certified Class I medical device, registered with the Italian Ministry of Health (RDM database no. 1763747, CDN Z12030682).

Reports & KPI

Sleep parameters* calculated
in the Sleepacta actigraphic report

*from internal data, european guidelines and patient age

WASO

Wake After Sleep Onset

Awakenings after falling asleep (in min)

TST

Total Sleep Time

Total sleep time (in min)

Sleep Efficiency

Sleep efficiency (in %)

SRI

Sleep Regularity Index

How regular is sleep

Workflow

The most advanced and reliable artificial intelligence algorithm for simplified sleep-wake cycle diagnostics

The Dormi platform is designed to be used by system integrators who wish to deliver Sleepacta’s certified actigraphic service in a third-party information system (a web application, a mobile app, a totem, etc.).

Partners wishing to integrate sleep into their own IT applications should equip themselves with “commercial” activity trackers – low-cost and comfortable to wear – supported by Sleepacta (e.g., FitBit) and previously initialized in Dormi.

The patient will wear the device for a predetermined number of days (3 to 7) without removing it.

At the end of the data collection session, the partner will invoke a Dormi API to request the execution of the actigraphic examination in the time interval of interest.

Following the request to perform an exam, the Dormi platform:

Will interact via API with the server of the wearable device manufacturer to retrieve and collect the “raw data” collected from the wearable device of interest, in the given time interval.
Will run the algorithm and produce the reports & KPIs of the actigraphic exam, returning them either in a standard JSON format, to allow customized republishing, or as static images, in the certified Sleepacta format.
Will notify the partner’s calling system of the execution of the exam so that it can retrieve the details.

API REST

The REST API layer

Dormi's REST APIs are authenticated, and allow, in addition to the execution of the algorithm, a number of management-type functionalities:

Census “partners” (parties in charge of managing patients)

Management of deliveries of activity trackers to partners, and from partners to patients

Patient’ census (with anonymisation of master data)

Activity trackers’ census

Auditing log of actigraphic examinations performed, with the possibility of retrieving the relevant raw data

Documentation

Output

The actigraph trace and the reference values of the monitored parameters

There are several views that you can have of the actigraphic examination. To make the tool easily interpretable, the data can be viewed in a report that can be customized.
In its basic form Sleepacta makes its algorithms available to its partners in as-a-service mode through an API layer. A web application is also available that uses the same API layer and exposes the same functions, providing additional reporting and control interfaces.

Customers and Partners

Scientifically validated technologies at the service of our customers

Sleepacta boasts partnerships and develops projects in collaboration with schools of excellence, research centers and university foundations of international standing, and applies its technologies in the field, in collaboration with hospitals.

To know more

hand-with-fitness-tracker-2021-06-01-17-30-34-utc@2x

Areas of application

Dormi, an algorithm easily integrated into all Healthtech systems

Based on a full-API approach and designed to be an as-a-service solution, Dormi technology can be easily integrated into existing software platforms, regardless of the field of specialization or business.

To know more

Contacts

Are you interested in integrating the Dormi algorithm into your diagnostic systems?

To know more about sleep

Well-being and health start with sleep, find out why

There is a direct correlation between sleep disorders and a number of diseases and improving sleep diagnostics can help treat them.

To know more