Hybrid database to store, query and analyze waveform data and annotations with an HTTP API.
| Real-time insertions | Big-data analytics (Hadoop...) | Fast get with filtering |
|---|---|---|
| ✓ | ✓ | ✓ |
This database originally aims to solve the problem of storing healthcare monitor waveforms while querying them or analyse them with Hadoop Hive.
This project is only a POC for now.
Immutable data:
- Raw waveforms (high frequency data, does not stores timestamps for each value, needs constant frequency data)
- Numerics (low frequency data, stores a value with a timestamp)
- Metadata such as source_id, type_id...
Mutable data:
- Annotation types
- Timestamp-based annotations
- Timerange-based annotations
Writes:
- Write waveforms, numerics
- Write annotations
Read (you can combine these options):
- Get data from date A to date B
- Get data where record_length >= 2hours
- Get data where bed_id=X, signal_type=ECG
- Get data where there are arythmia annotations
There are two types of data:
- Immutable: waveforms and numerics
- Editable/Expandable: metadata and annotations
Immutable data is directly stored in files that are not supposed to be editable. It must be really fast when deserializing them.
Editable/Expandable data is stored in an easily queryable store (sqlite/pgsql/mysql/... - must be compatible with sqlalchemy).
python3 api.py hostname port
An annotation type might be like 'bradycardia arrhythmia', 'start of record', 'end of record', 'heart R peak' or whatever you'd like
* GET <api-url>/annotations/types
-> returns all the annotations types
* GET <api-url>/annotations/types/<id>
-> returns the annotation type where id=<id>
* POST <api-url>/annotations/types (parameters=name:string)
-> returns the created annotation type with http code 201
* PUT <api-url>/annotations/types/<id> (parameters=name:string)
-> updates the annotation type where id=<id> and returns it
* DELETE <api-url>/annotations/types/<id>
-> deletes the annotation type where id=<id> and returns http code 204
* GET <api-url>/annotations/timestamp
-> returns all the annotations that are timestamp-based
* GET <api-url>/annotations/timestamp/<id>
-> returns the annotation where id=<id>
* POST <api-url>/annotations/timestamp (parameters=source_id:int, type_id:int, value:float, comment:string, timestamp_micros:int)
-> returns the created annotation with http code 201
* PUT <api-url>/annotations/timestamp/<id> (parameters=source_id:int, type_id:int, value:float, comment:string, timestamp_micros:int)
-> updates the annotation where id=<id> and returns it
* DELETE <api-url>/annotations/timestamp/<id>
-> deletes the annotation where id=<id> and returns http code 204
* GET <api-url>/annotations/timerange
-> returns all the annotations that are timerange-based
* GET <api-url>/annotations/timerange/<id>
-> returns the annotation where id=<id>
* POST <api-url>/annotations/timerange (parameters=source_id:int, type_id:int, values:float, comment:string, start_micros:int, end_micros:int)
-> returns the created annotation with http code 201
* PUT <api-url>/annotations/timerange/<id> (parameters=source_id:int, type_id:int, values:float, comment:string, start_micros:int, end_micros:int)
-> updates the annotation where id=<id> and returns it
* DELETE <api-url>/annotations/timerange/<id>
-> deletes the annotation where id=<id> and returns http code 204
* GET <api-url>/waveforms?lf=False&hf=True&start_micros=1525349279579912&end_micros=1525349290101982&source_id=7&type_id=2
-> returns all the high frequency (hf) data between 1525349279579912 and 1525349290101982 that has source_id=7 and type_id=2
* GET <api-url>/waveforms (parameters=lf_source_id:int, lf_type_id:int, lf_value:float, lf_timestamp_micros:int)
-> returns http code 201 if successful
* GET <api-url>/waveforms (parameters=hf_source_id:int, hf_type_id:int, hf_frequency:float, hf_start_micros:int, hf_values:list<float>)
-> returns http code 201 if successful