PD ISO TS 22077 — Health Informatics Medical Waveform Format (Parts 2 & 3: ECG and Long Term ECG)

🫀 PD ISO/TS 22077 is a BSI Published Document covering the medical waveform format in health informatics. Physiological waveforms such as ECG, EEG and blood pressure are core data for clinical diagnosis and monitoring, but because different manufacturers use different storage formats, waveform data has long been difficult to share and interoperate. This series provides a unified encoding framework for medical waveforms: Part 2 specifies the electrocardiography (ECG) waveform format and Part 3 specifies the long term electrocardiography waveform format, both prepared by BSI Technical Committee IST/35 (health informatics) and published in August 2015.

1. Scope and General Framework 📌

PD ISO/TS 22077 applies to the encoding and interchange of medical waveform data. Part 2 specifies the waveform encoding format for routine (resting) electrocardiography, covering sampling attributes, frame attributes and encoding representation. Part 3 specifies the waveform encoding format for long term electrocardiography (Holter), aimed at continuous recordings of 24 hours or longer, with higher demands on data compression and storage efficiency. The common goal of both parts is to provide a standard exchange language for ECG waveforms across devices and systems.

2. Encoding Format and Key Attributes 🔧

The medical waveform format encoding consists of two levels of description:

• Primary description: defines the overall structure of the waveform group, including number of channels, sampling frequency, bit depth, symbol system and compression method.
• Sampling attributes: define the physical dimension, scaling (gain/offset), units and time stamp of each sample, so that values can be reconstructed into real physiological signals.
• Frame attributes: define how waveform frames are organised, including frame length, frame interval, synchronisation information and annotation/event marking.

In addition, the format defines the representation of patient information, recording device information, lead configuration (e.g. standard 12-lead) and event markers (e.g. arrhythmia events, pacemaker markers), providing context for clinical interpretation and statistical analysis.

3. Special Requirements for Long Term ECG 🔩

Long term ECG (Part 3) targets continuous recordings of 24 hours or longer, whose data volume far exceeds routine resting ECG, so the format has special requirements:

• Efficient compression: support lossless or clinically acceptable lossy data compression to reduce storage and transmission burden.
• Long-term stability: the format should support accurate time stamps and drift correction over long continuous recordings.
• Event markers: support recorder/patient event-button markers, automatic arrhythmia event detection markers and annotations.
• Extensibility: allow extended channels (e.g. SpO₂, respiration) and custom metadata, adapting to the diverse needs of ambulatory monitoring.

In practice, a balance must be struck between compression ratio and clinical fidelity, ensuring algorithm validity in arrhythmia analysis, ST analysis and other clinical applications.

4. Interoperability and Engineering Application 🔍

• 🏥 Device interoperability: ECG recorders, monitors and EHR systems export/import waveforms in the standard format, enabling intra- and inter-hospital data sharing.
• 🌐 Remote monitoring: the standard format supports AI-assisted diagnosis and tele-consultation platforms, with a unified interface reducing integration cost.
• 🔬 Data research: the standardised format eases the construction of large ECG databases for algorithm training and clinical research.
• 🔄 Migration and archiving: archive waveform data in the standard format to avoid vendor lock-in and ensure long-term availability.

4b. Practical Guidance and Data Handling 🔧

Applying the medical waveform format in a clinical environment requires attention to several operational points. First, the sampling attributes must truly reflect the acquisition settings: the nominal sampling frequency of the recorder, the gain and offset applied to each channel, and the bit depth determine whether the stored values can be converted back into millivolt signals with diagnostic accuracy. Second, the frame attributes should be populated consistently, because the correlation of events, annotations and time stamps across channels is what allows automated algorithms and clinicians to interpret the record reliably. Third, when data from different devices are merged into a single archive or database, the metadata should be checked for consistency, since different recorders may use different sampling rates or lead configurations. Fourth, for long-term records, time synchronisation between the waveform stream and external event markers (patient diaries, pacemaker markers) is essential for correct interpretation. Fifth, data exchange should reference the standard version in use and document any vendor-specific extensions, so that the receiving system can handle them correctly or at least flag them.

For storage planning, an uncompressed 12-lead record can be large, so Part 3 of the series supports compression choices that must be declared in the primary description. Lossless compression is recommended for research and diagnostic archives, while lossy compression may be acceptable for monitoring and screening provided the clinical fidelity requirements are met and the processing is documented. Finally, privacy considerations apply throughout the lifecycle: de-identification of patient identifiers, access control and audit trails should be implemented before waveforms leave the hospital network, and retention policies should follow national regulations on health records.

In summary, the medical waveform format provides the encoding framework that turns raw physiological signals into interoperable, self-describing clinical data. Adopting it improves the reliability of remote monitoring, supports large-scale research and enables the safe reuse of waveform data across institutions and countries.

5. Frequently Asked Questions ❓