TL;DR
This paper introduces mobileSpiro, a low-cost, smartphone-connected spirometer designed for at-home asthma self-management. It tackles the challenge of inaccurate patient spirometry by integrating robust, automated error-detection algorithms directly into the Android application, enabling reliable daily monitoring without the need for a technician.
What Problem Does This Paper Address?
Effective asthma management hinges on regular monitoring of lung function, typically done with a spirometer. However, despite decades of clinical use, personalized, patient-operated spirometers are scarce, especially those that integrate with smartphones. A major hurdle is the difficulty patients face in performing the forced expiratory maneuver correctly. Errors like insufficient effort, coughing, or premature exhalation can invalidate readings, leading to false alarms or missed diagnoses. This lack of accessible, reliable self-monitoring tools forces many patients, particularly children with severe asthma, to rely on infrequent doctor visits, missing crucial early signs of worsening lung function.
Methods
The mobileSpiro system comprises three core components:
- Hardware: A custom, low-cost, portable spirometer featuring a pneumotachometer (Freescale MPXV7002) to measure airflow, an RC filter for noise reduction, and a Texas Instruments MSP430 microcontroller. This hardware connects wirelessly to an Android device via Bluetooth.
- Software API: A pure Java API designed for the Android platform, handling data collection from the sensor, calibration, and storage of spirometry maneuvers in a local SQLite database.
- Spirometry Validation Algorithms: This is the paper's key innovation. Real-time algorithms running on the smartphone analyze the collected data to detect common errors in the forced expiratory maneuver. These include:
- Cough Detection: Identifies aberrations in the flow-time curve that deviate from a single, sharp peak.
- Hesitation/Slow Start: Detects if the initial exhalation effort is not maximal, indicated by a less steep rise in volume over time or a late peak expiratory flow.
- Early Termination: Identifies maneuvers that end prematurely, either due to glottis closure (sudden flow drop to zero) or short-lived effort (flow dropping to zero before the required duration).
- Feedback System: Based on detected errors, the app provides simple, understandable coaching messages to guide the patient towards a better maneuver.
Datasets
The paper utilizes the CDC released NHANES III database for evaluating the accuracy of its error detection algorithms. This dataset contains spirometry data annotated by both technicians and computers, allowing for a benchmark comparison.
Key Insights and Ideas
- Democratizing Spirometry: The core idea is to make accurate spirometry accessible and affordable for everyday patients, shifting the burden of monitoring from infrequent clinical visits to daily self-assessment.
- Leveraging Smartphone Power: The project effectively utilizes the ubiquitous computing power of smartphones to perform complex data analysis and provide immediate feedback, reducing the need for specialized, expensive hardware.
- Automated Quality Control: The most significant insight is the development of robust, real-time algorithms that mimic a technician's role in validating spirometry maneuvers. This automated quality control is crucial for ensuring the reliability of data collected by non-experts.
- Patient Engagement through Feedback: Providing immediate, actionable feedback based on detected errors is key to improving patient technique and fostering engagement with their own health management.
Key Results
- Hardware Accuracy: The mobileSpiro hardware demonstrated an inter-device deviation of less than 8% compared to a commercial ISO13485-certified spirometer (Thor PC FlowMeter) across a range of flow rates.
- Error Detection Accuracy:
- Cough Detection: The algorithms successfully identified 95.5% of cough maneuvers from the NHANES III dataset, with a 15% false positive rate.
- Early Termination Detection: The algorithms achieved 74% detection accuracy for early termination maneuvers, with a 14% false positive rate.
- Cost-Effectiveness: The custom hardware component costs **less than $100** in large quantities, significantly lower than commercial spirometers priced over $500.
- Power Efficiency: The device can operate for up to 80 days on standard AA batteries with typical usage (twice daily, 10 minutes each).
Limitations and Future Work
- Hesitation Detection: The paper notes that validating hesitation detection algorithms was difficult due to inconsistent labeling in the NHANES III dataset.
- Future Work: The authors mention the potential for server-side implementations to allow for greater flexibility in patient data access and more sophisticated analysis. They also suggest integrating more advanced measures and other medical sensors to provide a more personalized health portrait. The current system is primarily focused on flow and volume data, with symptom logging as a secondary input.
Summary updated