.png?width=500&height=278&name=Customer%20Cases%20image%20(24).png)
Deep Breath: Turning Real-Time Ventilator Data into Actionable Insights for Hospitals
Deep Breath is now equipped with a scalable analytics service on Google Cloud, enabling them to offer hospitals enhanced ventilator care optimization.
monitored parameters
control settings
alarms per patient
The Goal
Deep Breath is on a mission to improve hospital care for patients who rely on ventilators. These devices send real-time data—every few seconds—to monitor a patient’s breathing. If oxygen levels drop too low or rise too high, or if the patient disconnects from the device, alarms are triggered. It’s critical that nurses respond quickly to avoid dangerous situations.
To help hospitals better understand and act on these alarm patterns, Deep Breathe needed reports to answer key questions:
- How many alarms occur in general? Per patient? Per type?
- How quickly do nurses respond to each type of alarm?
- What actions are taken to resolve the issue?
- Are some devices or patients triggering too many alarms?
They also wanted to preserve this high-frequency data (which would otherwise be lost after 3 months) and analyze it over time to support clinical studies and provide clear insights to hospital management and decision-makers.
The Challenge
Deep Breath already collects ventilator data from hospital-owned devices via a live data stream into their database. The dataset is rich and complex and they needed a solution that could:
- Translate technical data into clear, insightful dashboards
- Visualize key metrics like time to react, alarm frequency, and alarm type breakdown
About Deep Breathe
DeepBreath is based in Rotterdam, The Netherlands, provides an AI-driven decision support and remote monitoring system called Deep Breath, designed for hospital ICUs. Their mission is to optimize mechanical lung ventilation by processing real-time data from ventilators and patient monitoring systems. Using AI models trained on vast amounts of respiratory data, their technology aims to detect early signs of complications like ARDS and pneumonia, forecast adverse events, and provide treatment recommendations to clinicians. Ultimately, Deep Breath Tech seeks to improve patient outcomes, reduce mortality rates and ICU stays, and help increase hospital capacity through enhanced, data-driven ventilation management.
Start Your Success Story with Crystalloids
Partner with Crystalloids and Google Cloud to bring your project to life.
The Solution
Crystalloids was brought in to develop a scalable reporting layer using Looker and Google Cloud technologies. We started with a historical export from one hospital:
- 9 patients
- 765,000 alarms over several months
The underlying dataset was rich: more than 30 monitored parameters and 15 control settings from a Postgres database, representing the full complexity of breath and device behaviour.
From this dataset, we built an architecture that integrated BigQuery as the scalable analytics layer.
In Looker, we created a semantic layer using LookML and developed two user-friendly dashboards:
- Overview dashboards with filters by alarm type, patient, and time period
- Drill-downs into individual patient profiles and event timelines
- Visualizations for alarm frequency, time-to-response, and nurse actions
The Result
The result is a streamlined, intuitive reporting solution that makes complex clinical data understandable—even for non-technical stakeholders. Hospital staff and decision-makers can now:
- See which alarms are most frequent and how long they take to resolve
- Monitor nurse response time and care quality
- Identify problematic patterns at the patient or device level
- Support clinical studies with reliable long-term data
The dashboards now serve as a valuable asset that Deep Breath can offer to hospitals, enabling better care through data-driven decision-making. This successful Proof of Concept marks the beginning of Deep Breath's journey towards growth.
