Publications
Notice: These academic theses studied earlier prototypes and research concepts. They do not describe the intended purpose, features or performance of Spiroluft, which is under development and not CE-marked.
SPIROLUFT: A Next Generation Breath Analyser Utilising Simultaneous Spirometry & Pulse Oximetry
Date: November 2024
Säfström, Felix – Uppsala University, Disciplinary Domain of Science and Technology, Mathematics and Computer Science, Department of Information Technology.
Master’s thesis on an early Spiroluft prototype combining spirometry and pulse oximetry (pulse oximetry is not part of the current design).
Development of a Breath Monitoring System Using an STM32 Microcontroller and a Bluetooth Device with Connected MEMS-Based Sensors
Date: May 2024
Ravichandran, Vijayakanna – Uppsala University, Science and Technology Disciplinary Domain, Mathematical and Computer Science Section, Department of Information Technology.
Master’s thesis on a Bluetooth MEMS-based spirometer prototype.
Developing and Validation of a Safety-Critical Usability Framework for Dual-User Respiratory Health Systems
Date: June 2026
Chang, Ching Ho – Uppsala University, Disciplinary Domain of Humanities and Social Sciences, Faculty of Social Sciences, Department of Informatics and Media.
Abstract: The translation of breath analysis system from specialized hospitals to primary care and home self-monitoring introduces
a distinct “dual-user” challenge: a single interface must safely and effectively accommodates both expert clinicians and
lay patients. Existing mHealth usability heuristics fail to adequately address the safety-critical demands of this paradign
shift. To bridge this gap, this thesis introduces the Safety-Critical Interaction Dimensions (SCID), an 18-dimensional
heuristic framework grounded in human-computer interaction (HCI), respiratory clinical practice, and medical device
standards. The framework was developed through a rigorous six-phase methodology including regulatory gap analysis,
HCI expert consultation and clinician validation. The framework achieved a 95% item endorsement rate with strong interrater agreement (Gwet’s AC2 = 0.781). The SCID framework was then applied via heuristic evaluation to Spiroluft®, an
advanced portable breath analysis system. Five evaluators identified 79 unique usability issues across 16 of 18 dimensions,
Notably, the highest-severity issues clustered around forced expiratory maneuver, where interface failure directly threatens
measurement validity and patient safety. Ultimately, SCID offers a regulation-compliant, validated framework for ensuring
user safety in digital repiratoy systems, whether deployed by clinicians in primary care or by patients at home.
AI-powered respiratory health monitoring with predictive analytics and personalized insights
Date: August 2026
Prabhakaran Nair, Akhila – Uppsala University, Disciplinary Domain of Humanities and Social Sciences, Faculty of Social Sciences, Department of Informatics and Media
Abstract: Chronic obstructive pulmonary disease (COPD) often remains undiagnosed until significant lung damage has occurred. This research explores how machine learning can improve COPD classification and identify individuals at risk before diagnosis.
Using over 1.7 million patient records from US and Swedish population datasets, the study analysed spirometry, demographics, smoking history, and longitudinal health data. Advanced algorithms, including Random Forest and XGBoost, were trained using conventional spirometry indices and novel flow-volume curve features. A model retrained on Swedish clinical data achieved an AUC of 0.954, with 90.9% sensitivity and 91.2% specificity. Importantly, machine learning models predicted future COPD development up to three years before a recorded diagnosis, achieving AUCs of 0.72–0.73. The findings highlight the potential of AI to extract valuable respiratory insights beyond conventional spirometry ratios. The research also demonstrates the importance of adapting models to local patient populations for optimal performance. These results support further development of AI-enabled digital respiratory health platforms such as Spiroluft. Prospective validation across diverse patient populations is the next essential step towards responsible clinical implementation.
Note: Spiroluft contains no AI or prediction function.