ppg listlengths 200 5.39k | bpm int64 46 157 | file_id stringlengths 37 37 |
|---|---|---|
[
{
"b": 1.6243,
"g": 0.186,
"r": 219.198,
"t": 0,
"x": 2.5857,
"y": 1.9632,
"z": 10.07
},
{
"b": 1.6243,
"g": 0.186,
"r": 219.198,
"t": 0.1923,
"x": 2.7749,
"y": 2.2051,
"z": 9.584
},
{
"b": 1.8883,
"g": 0.2041,
"r": 221.766,
"t": 0.2307... | 97 | 000025e41b787c4874e1ed067769379c.json |
[{"b":151.0959,"g":164.0891,"r":178.2588,"t":0.0,"x":1.7097,"y":4.7339,"z":7.5622},{"b":149.1445,"g"(...TRUNCATED) | 62 | 00003261ef4a262b98a48846e84bd114.json |
[{"b":196.6468,"g":207.2554,"r":210.5658,"t":0.0,"x":0.984,"y":9.1961,"z":3.5721},{"b":153.8932,"g":(...TRUNCATED) | 86 | 000033d29000749546b9533676801e4b.json |
[{"b":124.7958,"g":208.7353,"r":254.288,"t":0.0,"x":-2.052,"y":2.0951,"z":9.026},{"b":127.6555,"g":2(...TRUNCATED) | 80 | 000042384e88913c224c12937f4e4657.json |
[{"b":57.059,"g":56.2888,"r":51.3546,"t":0.0,"x":2.195,"y":7.422,"z":5.515},{"b":51.7655,"g":47.4781(...TRUNCATED) | 74 | 00004c721a444ad4a1acefba10a56535.json |
[{"b":132.8337,"g":130.6306,"r":161.1535,"t":0.0,"x":0.842,"y":7.605,"z":6.269},{"b":133.0278,"g":13(...TRUNCATED) | 93 | 00005162d6b6b092a774f837811eba32.json |
[{"b":175.6093,"g":187.9821,"r":194.5369,"t":0.0,"x":-8.945,"y":1.245,"z":4.175},{"b":205.2213,"g":2(...TRUNCATED) | 80 | 00005b45e3a2704fd27bcd8fee51e857.json |
[{"b":130.6393,"g":139.3555,"r":136.9969,"t":0.0,"x":0.8021,"y":9.495,"z":-1.193},{"b":127.9142,"g":(...TRUNCATED) | 65 | 0000761a119b0b0958dca7a979301b01.json |
[{"b":219.3102,"g":227.3342,"r":231.7889,"t":0.0,"x":7.3666,"y":6.2952,"z":-3.3147},{"b":219.7424,"g(...TRUNCATED) | 110 | 000092f2a6be461862ae10b5dfd24706.json |
[{"b":10.2948,"g":27.0665,"r":131.374,"t":0.0,"x":-2.0578,"y":9.3937,"z":-1.6346},{"b":2.1243,"g":1.(...TRUNCATED) | 64 | 0000c356ac70bf44d050d335c2417326.json |
Spandan
A Large Photoplethysmography (PPG) Signal Dataset of 1 Million+ Indian Subjects
In Sanskrit, "Spandan" (स्पन्दन - spandana) represents one of the most fundamental aspects of existence - the rhythmic pulsation that permeates all life. Derived from the root verb "spand" (स्पन्द), meaning "to throb" or "to pulsate," it beautifully captures the essence of the heartbeat.
Dataset Overview
Spandan is an extensive repository containing over 1 million Photoplethysmography (PPG) signals captured across diverse mobile devices throughout India. This comprehensive dataset represents a curated subset of data collected through Eka Care's Personal Health Record (PHR) mobile application on Android and iOS platforms. In addition to raw PPG signals, the dataset includes heart rate estimations derived from Eka Care's proprietary algorithm.
- Curated and Shared by: Eka.Care
- License: CC-BY-SA-4.0
Data Acquisition
The PPG signals were captured using mobile device cameras through the following Personal Health Record applications:
- Applications: Eka Care PHR mobile applications
- Publication: Smartphone-based health monitoring in India: Data collection and evaluation for pulse rate estimation
Data Structure and Parameters
Each file contains the following components:
PPG Signal Data
- r: Mean value of the red channel from RGB video capture
- g: Mean value of the green channel from RGB video capture
- b: Mean value of the blue channel from RGB video capture
- t: Time elapsed from signal capture initiation
Motion Data
- x: x-axis acceleration recorded by mobile gyroscope
- y: y-axis acceleration recorded by mobile gyroscope
- z: z-axis acceleration recorded by mobile gyroscope
Heart Rate Estimation
- bpm: Estimated heart rate in beats per minute
Note: For computing the mean of each RGB channel, 75% of the middle region of the screen was considered.
Potential Applications
- Pre-training and SFT of foundation models for physiological signal processing
- Developing non-invasive blood pressure and heart rate variability estimation
- Establishing India-specific cardiovascular baselines and reference ranges
- Developing motion artifact compensation techniques for real-world conditions
- Benchmarking frequency and time-domain analysis methods
- Developing educational resources for biomedical signal processing
- Comparing traditional signal processing against deep learning approaches
Dataset Creation Methodology
Data Collection and Processing
The shared corpus is a carefully selected subset of data collected through Eka Care's PHR applications. To ensure representativeness, unique users were randomly selected for inclusion in the dataset, ensuring diversity across age, gender, geographic location, and heart rate values.
The PPG signals included in this dataset are raw signals captured directly from mobile devices (represented as mean pixel values) without any post-processing.
Privacy Considerations
This dataset contains absolutely no personally identifiable information (PII) and is completely anonymized.
Citation
Shetty, Achal; Narasimhamurthy, Sanjana S; Nataraj, KS; Prabhu, Srilakshmi M; Jagadeesh, Neha; Katre, Kunal; Kumar, Sumit; Kapoor, Neelesh; Haladi, Sudhir P; Gulati, Sankalp. Smartphone-based health monitoring in India: Data collection and evaluation for pulse rate estimation. Journal of Family Medicine and Primary Care 14(1):p 348-355, January 2025. | DOI: 10.4103/jfmpc.jfmpc_1257_24
Contact Information
For inquiries, please email with the subject line "Spandan enquiry" to: sankalp [at] eka.care
- Downloads last month
- 53