# Health Indices Dataset from Clinical Study A - TOLIFE Project

## Overview

This dataset contains sensor-derived health indices collected as part of the TOLIFE project (Combining Artificial Intelligence and smart sensing TOward better management and improved quality of LIFE in COPD, Grant Agreement No. 101057103), which clinically validates an artificial intelligence solution to process daily life patient data captured by unobtrusive sensors for COPD (Chronic Obstructive Pulmonary Disease) monitoring. The data represents continuous monitoring of health parameters from participants of Clinical Study A (CSA) across three European locations (Germany, Spain and Italy) between May 2024 and April 2026.

In addition to the sensor-derived health indices, the dataset includes clinically annotated information such as COPD exacerbation events, classified by severity, and detailed clinical and demographic data collected during scheduled clinical visits (timepoints T1, T2, T3, T4, T5) with healthcare professionals. These additional sources provide important context for interpreting the sensor data, capturing aspects such as disease progression, patient-reported outcomes and comorbidities.

This is version 2.0 of the dataset. It supersedes version 1.0 (July 2025), which covered the first year of monitoring for 74 participants. Version 2.0 covers the complete CSA monitoring period for 86 participants, with clinical visits and exacerbation records updated accordingly. See "Version History" for details.

## Dataset Structure

The dataset is organized in a hierarchical structure:

```
index/
├── kits.csv                     # One row per participant: country, patient ID, monitoring period, counts
├── clinical_indices_all.csv     # Clinical data of all CSA participants in a single table (with Kit-ID)
├── exacerbations_all.csv        # All exacerbation events in a single table (with Kit-id and country)
├── GERMANY/
│   └── GER-XXX/                 # Individual participant directories
│       ├── EI.csv               # Environmental Index
│       ├── SHRI.csv             # Sleeping Heart Rate Index
│       ├── SII.csv              # Sound Intensity Index
│       ├── MI.csv               # Mobility Index
│       ├── BRI.csv              # Breathing Rate Index
│       ├── SQI.csv              # Sleep Quality Index
│       ├── PRVI.csv             # PPG Pulse Rate Variability Index
│       ├── PEI.csv              # Pulmonary Efficiency Index
│       ├── CAT.csv              # Daily estimated COPD Assessment Test score
│       ├── CCQ.csv              # Daily estimated Clinical COPD Questionnaire score
│       ├── mMRC.csv             # Daily estimated mMRC dyspnea score
│       ├── clinical_indices.csv # Clinical and demographic data
│       └── exacerbations.csv    # Exacerbations data (when present)
├── SPAIN/
│   └── SPA-XXX/                 # Same file set as above
└── ITALY/
    └── ITA-XXX/                 # Same file set as above
```

The two `*_all.csv` tables also include a small number of CSA participants for whom clinical records exist but no sensor data was collected (these participants have no directory of their own).

## Data Collection Period

- **Start Date**: May 2024
- **Data Coverage**: May 2024 to April 2026
- **Collection Frequency**: Daily monitoring; every index is reported once per day or per night, except the environmental and sound indices, which are reported twice a day (see below)
- **Status**: This release covers the complete monitoring period of Clinical Study A

Participants were enrolled progressively between May 2024 and September 2025 and followed for about 12 months each, so individual monitoring periods differ. The exact first and last day of monitoring of each participant is reported in `kits.csv`.

## Participant Information

### Germany
- **Total Participants**: 43 monitored participants
- **Kit ID Format**: GER-XXX (3-digit sequential numbering)

### Spain
- **Total Participants**: 41 monitored participants
- **Kit ID Format**: SPA-XXX (3-digit sequential numbering)

### Italy
- **Total Participants**: 2 monitored participants
- **Kit ID Format**: ITA-XXX (3-digit sequential numbering)

Participant identifiers are kit IDs. Each participant also has a numeric Patient-ID used in the clinical files; the correspondence is given in `kits.csv`.

## Clinical Indices Overview

| File Name | Index Type | Full Name | Resolution | Description |
|-----------|------------|-----------|------------|-------------|
| `EI.csv` | Environmental Index | Environmental Index | 2 per day | Environmental conditions (temperature, humidity, air quality, light) |
| `SHRI.csv` | Sleeping Heart Rate Index | Sleeping Heart Rate Index | 1 per night | Cardiovascular parameters during sleep |
| `SII.csv` | Sound Intensity Index | Sound Intensity Index | 2 per day | Audio-based health monitoring and sound measurements |
| `MI.csv` | Mobility Index | Mobility Index | 1 per day | Physical activity and movement patterns |
| `BRI.csv` | Breathing Rate Index | Breathing Rate Index | 1 per night | Respiratory frequency and breathing patterns |
| `SQI.csv` | Sleep Quality Index | Sleep Quality Index | 1 per night | Sleep efficiency and quality indicators |
| `PRVI.csv` | Pulse Rate Variability Index | pulse_rate_Variability_index | 1 per day | PPG smartwatch data including heart rate variability metrics |
| `PEI.csv` | Pulmonary Efficiency Index | Spirometer pulmonary_efficiency_index | 1 per day | Spirometry measurements including FEV1 and PEF parameters |
| `CAT.csv` | Estimated CAT | Daily estimated COPD Assessment Test | 1 per day | Model-based daily estimate of the CAT symptom class (< 10 / ≥ 10) from the sensor-derived indices |
| `CCQ.csv` | Estimated CCQ | Daily estimated Clinical COPD Questionnaire | 1 per day | Model-based daily estimate of the CCQ symptom class (< 1 / ≥ 1) from the sensor-derived indices |
| `mMRC.csv` | Estimated mMRC | Daily estimated mMRC dyspnea scale | 1 per day | Model-based daily estimate of the mMRC grade from the sensor-derived indices |
| `clinical_indices.csv` | Clinical Data | Clinical Indices | per visit (T1-T5) | Patient demographics, clinical measurements, and questionnaire scores |
| `exacerbations.csv` | Exacerbations Data | Exacerbations Data | per event | Clinical exacerbation events with severity classification |

## Data Types and Parameters

### 1. Environmental Index (`EI.csv`)
Environmental conditions measured in the participant's sleep environment, two records per day: one record for the night-time period, timestamped at its start (around 18:00 UTC, 19:00 UTC in winter), and one for the daytime period, timestamped at its start (around 04:00 UTC, 05:00 UTC in winter):

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| Mean_Temp | Average temperature | °C | float |
| Std_Temp | Temperature standard deviation | °C | float |
| Max_Temp | Maximum temperature | °C | float |
| Min_Temp | Minimum temperature | °C | float |
| Mean_Hum | Average humidity | % | float |
| Std_Hum | Humidity standard deviation | % | float |
| Max_Hum | Maximum humidity | % | float |
| Min_Hum | Minimum humidity | % | float |
| Mean_AQI | Average Air Quality Index | AQI scale (0-500) | float |
| Std_AQI | AQI standard deviation | AQI scale (0-500) | float |
| Max_AQI | Maximum AQI | AQI scale (0-500) | float |
| Min_AQI | Minimum AQI | AQI scale (0-500) | float |
| Mean_Lux | Average light intensity | lux | float |
| Std_Lux | Light intensity standard deviation | lux | float |
| Max_Lux | Maximum light intensity | lux | float |
| Min_Lux | Minimum light intensity | lux | float |
| index_validity | Data quality flag | 0/1 | integer |
| data_validity | Data availability flag | 0/1 | integer |

### 2. Sleeping Heart Rate Index (`SHRI.csv`)
Cardiovascular monitoring parameters during sleep, one record per night:

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| hr_mean | Average heart rate | bpm | float |
| hr_std | Heart rate standard deviation | bpm | float |
| hr_max | Maximum heart rate | bpm | float |
| hr_min | Minimum heart rate | bpm | float |
| index_validity | Data quality flag | 0/1 | integer |
| data_validity | Data availability flag | 0/1 | integer |

### 3. Sound Intensity Index (`SII.csv`)
Audio-based health monitoring and sound intensity measurements, two records per day: one record for the night-time period, timestamped at its start (around 18:00 UTC, 19:00 UTC in winter), and one for the daytime period, timestamped at its start (around 04:00 UTC, 05:00 UTC in winter):

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| Mean_SI | Average sound intensity | dimensionless | float |
| Std_SI | Sound intensity standard deviation | dimensionless | float |
| Max_SI | Maximum sound intensity | dimensionless | float |
| Min_SI | Minimum sound intensity | dimensionless | float |
| index_validity | Data quality flag | 0/1 | integer |
| data_validity | Data availability flag | 0/1 | integer |

### 4. Mobility Index (`MI.csv`)
Physical activity and movement indices, one record per day:

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| gs_mean | Average gait speed | m/s | float |
| gs_std | Gait speed standard deviation | m/s | float |
| sl_mean | Average step length | m | float |
| sl_std | Step length standard deviation | m | float |
| smwd_mean | Average six-minute walk distance | meters | float |
| smwd_std | Six-minute walk distance std dev | meters | float |
| walked_time | Total walking time | minutes | float |
| walked_dist | Total distance walked | meters | float |
| index_validity | Data quality flag | 0/1 | integer |
| data_validity | Data availability flag | 0/1 | integer |

### 5. Breathing Rate Index (`BRI.csv`)
Breathing pattern monitoring and respiratory frequency analysis during sleep phase, one record per night:

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| rf_mean | Average respiratory frequency | breaths/min | float |
| rf_std | Respiratory frequency standard deviation | breaths/min | float |
| rf_max | Maximum respiratory frequency | breaths/min | float |
| rf_min | Minimum respiratory frequency | breaths/min | float |
| index_validity | Data quality flag | 0/1 | integer |
| data_validity | Data availability flag | 0/1 | integer |

### 6. Sleep Quality Index (`SQI.csv`)
Sleep pattern and quality metrics, one record per night:

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| sleep_duration | Total sleep duration | hours | float |
| num_movements | Number of movements during sleep | count | integer |
| WASO | Wake After Sleep Onset | hours | float |
| SE | Sleep Efficiency | % | float |
| index_validity | Data quality flag | 0/1 | integer |
| data_validity | Data availability flag | 0/1 | integer |

### 7. Pulmonary Efficiency Index (`PEI.csv`)
Spirometry measurements including FEV1 and PEF parameters, one record per day. The timestamp of a record with a measurement is the time of the measurement session; days without any measurement are placed at 12:00 UTC. Because the daily window starts at 20:00 UTC of the previous day, a measurement taken after 20:00 UTC belongs to the following day while its timestamp keeps the date of the measurement, so two consecutive days can share the same calendar date:

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| mean_FEV1 | Average Forced Expiratory Volume in 1 second | cL | float |
| range_FEV1_low | FEV1 range lower bound | cL | float |
| range_FEV1_high | FEV1 range upper bound | cL | float |
| mean_PEF | Average Peak Expiratory Flow | cL/min | float |
| range_PEF_low | PEF range lower bound | cL/min | float |
| range_PEF_high | PEF range upper bound | cL/min | float |
| HR | Heart rate | bpm | float |
| SpO2 | Blood oxygen saturation | % | float |
| data_validity | Data availability flag | 0/1 | integer |
| index_validity | Data quality flag | 0/1 | integer |

### 8. PPG Pulse Rate Variability Index (`PRVI.csv`)
PPG smartwatch data including heart rate variability metrics, one record per day. The timestamp is not a fixed time of day: it depends on the smartwatch recording of that day and is mostly 12:00 UTC for days without a valid index. As for PEI, a record that starts after 20:00 UTC belongs to the following day while its timestamp keeps the date of the recording, so two consecutive days can share the same calendar date:

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Timestamp | UTC timestamp in ISO8601 format | YYYY-MM-DDTHH:MM:SSZ | datetime |
| mean_HR | Average heart rate | bpm | float |
| range_HR_low | Heart rate range lower bound | bpm | float |
| range_HR_high | Heart rate range upper bound | bpm | float |
| std_RR | Standard deviation of RR intervals | ms | float |
| range_stdRR_low | StdRR range lower bound | ms | float |
| range_stdRR_high | StdRR range upper bound | ms | float |
| RMSSD | Root Mean Square of Successive Differences | ms | float |
| range_RMSSD_low | RMSSD range lower bound | ms | float |
| range_RMSSD_high | RMSSD range upper bound | ms | float |
| nRMSSD | Normalized RMSSD | ms/bpm | float |
| range_nRMSSD_low | nRMSSD range lower bound | ms/bpm | float |
| range_nRMSSD_high | nRMSSD range upper bound | ms/bpm | float |
| data_validity | Data availability flag | 0/1 | integer |
| index_validity | Data quality flag | 0/1 | integer |

### 9. Daily Estimated Clinical Scores (`CAT.csv`, `CCQ.csv`, `mMRC.csv`)
Daily estimates of three standard COPD clinical scores, produced by models trained on Clinical Study A data (methods described in TOLIFE Deliverable D2.2). They are model outputs computed from sensor-derived indices only, not questionnaires filled in by the participant; the questionnaire scores collected at the clinical visits are in `clinical_indices.csv`. Each file has one record per day.

The estimate for day D uses the average of the valid values (`index_validity` = 1) of the following indices over the 7 days before D (D excluded):

- from `PEI.csv`: mean FEV1 and SpO2;
- from `PRVI.csv`: mean heart rate, stdRR and RMSSD;
- from `MI.csv`: gait speed and six-minute walk distance;
- heart rate and nRMSSD at rest, from the cardio-mobility analysis that combines smartwatch and mobility data (not distributed as a separate file).

| File | Column | Model | Values | Data Type |
|------|--------|-------|--------|-----------|
| `CAT.csv` | CAT | Linear discriminant analysis on the 9 predictors above | 0 = CAT < 10 (few symptoms), 1 = CAT ≥ 10 (many symptoms), following the GOLD grouping | integer |
| `CCQ.csv` | CCQ | Linear discriminant analysis on the 9 predictors above | 0 = CCQ < 1 (few symptoms), 1 = CCQ ≥ 1 (many symptoms), following the GOLD grouping | integer |
| `mMRC.csv` | mMRC | Cumulative link model (ordinal logistic regression) on mean FEV1, SpO2, gait speed, heart rate at rest and nRMSSD at rest | 1-5, same coding as `clinical_indices.csv`: 1 = mMRC grade 0 (no dyspnea) to 5 = grade 4 | integer |

All three files also contain `Timestamp` (UTC, ISO8601, midnight of day D), `index_validity` and `data_validity`. An estimate is computed when at most 3 of the 9 CAT/CCQ predictors, or 2 of the 5 mMRC predictors, are missing in the 7-day window; missing predictors are replaced by the training-set mean. Otherwise the record has value 0 with `index_validity` = 0 and `data_validity` = 0 and must be ignored.

### 10. Exacerbations Data (`exacerbations.csv`)
Clinical exacerbation events with severity classification:

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Date | Date of the exacerbation event | YYYY-MM-DD | date |
| PatientId | Patient identification number | - | integer |
| Kit-id | Participant kit identifier | - | string |
| Type | Exacerbation severity level | Moderate/Severe | string |

**Note**: The `exacerbations.csv` file is only present for participants who experienced clinical exacerbation events during the monitoring period. If this file is absent from a participant's directory, no exacerbations were recorded for that participant. The dataset-level `exacerbations_all.csv` contains the same records for all participants plus a `Country` column.

### 11. Clinical Indices Data (`clinical_indices.csv`)
Patient demographic information, clinical measurements, and standardized questionnaire scores collected at the five study visits (T1 at enrolment, then approximately every 3 months up to T5 at 12 months):

| Column | Description | Unit | Data Type |
|--------|-------------|------|-----------|
| Patient-ID | Patient identification number | - | integer |
| Timepoint | Study visit timepoint | T1-T5 | string |
| Visit Date | Date of clinical visit | YYYY-MM-DD | date |
| Year of Birth | Patient birth year | - | integer |
| Sex | Patient biological sex | Male/Female | string |
| Height (cm) | Patient height | cm | float |
| Weight (kg) | Patient weight, measured at T1, T3 and T5 | kg | float |
| FEV1 (L) | Forced Expiratory Volume in 1 second | L | float |
| FEV1_pred (%) | FEV1 percentage of predicted | % | float |
| FVC (L) | Forced Vital Capacity | L | float |
| FVC_pred (%) | FVC percentage of predicted | % | float |
| Lung Ratio | FEV1/FVC ratio | - | float |
| DLCO | Diffusing capacity of the lungs for carbon monoxide | mmol/(min*kPa) | float |
| DLCO_pred (%) | DLCO percentage of predicted | % | float |
| CAT Score | COPD Assessment Test score | 0-40 | integer |
| mMRC Score | Modified Medical Research Council dyspnea scale, eCRF coding (1 = grade 0, no dyspnea; 5 = grade 4) | 1-5 | integer |
| CCQ Total Score | Clinical COPD Questionnaire total score | 0-6 | float |
| CCQ Symptom | CCQ symptom domain score | 0-6 | float |
| CCQ Functional | CCQ functional domain score | 0-6 | float |
| CCQ Mental | CCQ mental domain score | 0-6 | float |
| Systolic BP (mmHg) | Systolic blood pressure | mmHg | integer |
| Diastolic BP (mmHg) | Diastolic blood pressure | mmHg | integer |
| Heart Rate (bpm) | Resting heart rate | bpm | integer |
| Oxygen Saturation (%) | Blood oxygen saturation | % | integer |
| EQ5D Mobility | EuroQol-5D mobility dimension | 1-5 | integer |
| EQ5D Self Care | EuroQol-5D self-care dimension | 1-5 | integer |
| EQ5D Usual Activities | EuroQol-5D usual activities dimension | 1-5 | integer |
| EQ5D Pain/Discomfort | EuroQol-5D pain/discomfort dimension | 1-5 | integer |
| EQ5D Anxiety/Depression | EuroQol-5D anxiety/depression dimension | 1-5 | integer |
| EQ5D Health Score | EuroQol-5D visual analogue scale | 0-100 | integer |
| Anxiety HADS | Hospital Anxiety and Depression Scale - Anxiety | 0-21 | integer |
| Depression HADS | Hospital Anxiety and Depression Scale - Depression | 0-21 | integer |
| Total PSQI Score | Pittsburgh Sleep Quality Index total score | 0-21 | integer |
| Sleep Efficiency (%) | Sleep efficiency percentage (PSQI) | % | float |
| Six-Minute Walk Distance (m) | Six-minute walk test distance | meters | float |

**Note**: Each file always contains five rows (T1 to T5). Rows of visits that did not take place, or had not taken place at the time of the data export, contain only Patient-ID and Timepoint. Empty cells in the other rows indicate measurements that were not collected or not applicable at that timepoint (for example, full spirometry and DLCO are assessed at T1, T3 and T5 only). The dataset-level `clinical_indices_all.csv` contains the same records for all participants plus `Kit-ID` and `Country` columns.

### 12. Participant Manifest (`kits.csv`)

| Column | Description |
|--------|-------------|
| Kit-ID | Participant kit identifier |
| Country | GERMANY, SPAIN or ITALY |
| Patient-ID | Patient identification number used in the clinical files |
| first_day | First day on which sensor data were recorded (YYYY-MM-DD) |
| last_day | Last day on which sensor data were recorded (YYYY-MM-DD) |
| n_days | Number of days on which sensor data were recorded |
| n_visits | Number of clinical visits with a recorded date |
| n_exacerbations | Number of recorded exacerbation events |

A day counts when at least one record of the eight sensor indices (EI, SHRI, SII, MI, BRI, SQI, PEI, PRVI) has `data_validity` = 1. Days are aggregation days, as described under "Timestamp Standardization": the day to which the pipeline assigns a record, which for evening records is the day after the calendar date of the timestamp. `first_day` can differ from the date of the T1 visit in `clinical_indices.csv`, because the collection of sensor data did not always start on the day of the enrolment visit. The daily estimated scores are not counted, because their files contain a record for every day of the study period, also when no sensor data were available.

## Data Quality Indicators

Each measurement includes two quality flags:

- **index_validity**: Indicates whether the calculated health index is valid (1) or invalid (0)
- **data_validity**: Indicates whether the raw sensor data was available (1) or missing (0)

Only records with `index_validity = 1` should be used for analysis. Records with `data_validity = 0` are kept so that periods without data can be identified; their measurement values are 0 (empty in `PRVI.csv`). Records with `index_validity = 0` and `data_validity = 1` had data that did not produce a valid index; in `MI.csv` and `PEI.csv` some of them still contain partial values, which should not be used.

## Data Processing

### Timestamp Standardization
All timestamps are expressed in UTC in ISO8601 format (`YYYY-MM-DDTHH:MM:SSZ`, with milliseconds for some indices) for consistency across data collection sites and local timezone differences. Each record keeps the timestamp of its own data. The pipeline builds the record(s) of day D from a 24-hour window running from 20:00 of day D-1 to 20:00 of day D: local time for the environmental and sound indices, UTC for the mobility, smartwatch and spirometer indices, while the night-time indices cover the night inside this window. A record measured in the evening of day D-1 therefore belongs to day D even though its timestamp carries the date of D-1, and it can share that calendar date with the record of day D-1 (see `PEI.csv` and `PRVI.csv`). Night-time indices have one record per night. In `SHRI.csv` and `BRI.csv` the timestamp is local midnight at the start of the calendar day on which the night begins, converted to UTC: for example, the night from 25 to 26 July 2024 at a German site is labelled `2024-07-24T22:00:00Z`. In `SQI.csv` the same night is labelled with 00:00 UTC of the day on which it ends (`2024-07-26T00:00:00.000Z`). The timestamp column is named `Timestamp` in every index file.

### Data Consolidation
Individual daily files for each index have been merged into single CSV files per participant kit, maintaining chronological order and removing duplicate entries. Clinical and demographic data have been extracted from the study electronic case report form and the clinical partners' visit records and harmonised into a single table per participant.

## Technical Specifications

### File Format
- **Format**: CSV (Comma-Separated Values)
- **Encoding**: UTF-8
- **Delimiter**: Comma (,)
- **Decimal Separator**: Period (.)
- **Missing Values**: Empty cells

### Data Volume
- **Total Participants**: 86 (43 German + 41 Spanish + 2 Italian)
- **Individual Index Files**: 946 files (11 indices × 86 participants)
- **Clinical Data Files**: 86 files (1 per participant)
- **Exacerbation Files**: Variable (only for participants with recorded exacerbations)
- **Collection Duration**: 24 months (May 2024 - April 2026)

## Ethical Considerations

This study was conducted in accordance with ethical guidelines for clinical research:
- Ethics committee approval obtained from participating institutions
- Written informed consent obtained from all participants
- Data has been anonymized with participant identifiers replaced by kit IDs
- Privacy protection measures implemented throughout data collection and processing
- Participants had the right to withdraw from the study at any time

## Associated Publications

Further details about the TOLIFE project methodology, sensor technologies, and analysis approaches can be found in the following publications:

1. Zanoletti, M., Bufano, P., Bossi, F., Di Rienzo, F., Marinai, C., Rho, G., Vallati, C., Carbonaro, N., Greco, A., Laurino, M., & Tognetti, A. (2024). Combining Different Wearable Devices to Assess Gait Speed in Real-World Settings. *Sensors*, 24(10), 3205. https://doi.org/10.3390/s24103205

2. Marinai, C., Arcarisi, L., Bossi, F., Bufano, P., Di Rienzo, F., Melissa, E., Rho, G., Zanoletti, M., Greco, A., Laurino, M., Vallati, C., Carbonaro, N., & Tognetti, A. (2024). Smart Mattress Cover for Unobtrusive Monitoring of Sleep-Quality Correlates in Real-Life. In *2024 IEEE SENSORS* (pp. 1-4). IEEE. https://doi.org/10.1109/SENSORS60989.2024.10784486

3. Rho, G., Carbonaro, N., Laurino, M., Tognetti, A., & Greco, A. (2024). Estimating Heart Rate Variability from Wrist-Worn Photoplethysmography Devices in Daily Activities: A Preliminary Convolutional Denoising Autoencoder Approach. In *2024 IEEE International Conference on Metrology for eXtended Reality, Artificial Intelligence and Neural Engineering (MetroXRAINE)* (pp. 207-212). IEEE. https://doi.org/10.1109/MetroXRAINE62247.2024.10796284

4. Carbonaro, N., Laurino, M., Greco, A., Marinai, C., Giannetti, F., Righetti, F., Di Rienzo, F., Rho, G., Arcarisi, L., Zanoletti, M., Bufano, P., Tesconi, M., Sgambelluri, N., Menicucci, D., Vallati, C., & Tognetti, A. (2024). Smart Sensors for Daily-Life Data Collection Toward Precision and Personalized Medicine: The TOLIFE Project Approach. In A. Badnjević & L. Gurbeta Pokvić (Eds.), *MEDICON'23 and CMBEBIH'23* (pp. 783-794). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-49062-0_82

5. Di Rienzo, F., Righetti, F., Laurino, M., Greco, A., Marinai, C., Di Mambro, I., Melissa, E., Carbonaro, N., Bossi, F., Rho, G., Arcarisi, L., Zanoletti, M., Bufano, P., Tognetti, A., & Vallati, C. (2024). Using Multiple Devices for Patient Monitoring in Clinical Studies: The TOLIFE Experience. In *2024 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)* (pp. 148-153). IEEE. https://doi.org/10.1109/PerComWorkshops59983.2024.10502767

6. Zanoletti, M., Bufano, P., Bossi, F., Di Rienzo, F., Marinai, C., Rho, G., Melissa, E., Vallati, C., Carbonaro, N., Greco, A., Tognetti, A., & Laurino, M. (2025). Predicting Six-Minute-Walking-Distance in COPD Patients From Wearable Devices in Real-World Setting. In *2025 47th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)* (pp. 1-7). IEEE. https://doi.org/10.1109/EMBC58623.2025.11252978

7. G. Rho, N. Carbonaro, M. Laurino, A. Tognetti, and A. Greco, "A Convolutional Denoising Autoencoder approach for estimating Heart Rate Variability from wrist photoplethysmography devices in daily activities," under review, Biomedical Signal Processing and Control, Elsevier.

8. M. Zanoletti, C. Vallati, N. Carbonaro, A. Greco, A. Tognetti, and Marco Laurino, "Real-World Gait Speed Estimation: an AI-based Approach for Adaptive Wearable Devices Integration," under review, Journal of Biomedical and Health Informatics, IEEE

9. Marinai, C., Melissa, E., Di Rienzo, F., Vallati, C., Laurino, M., Tognetti, A. & Carbonaro, N., (2025). "The Smart Mattress Cover System: a novel textile-based system for unobtrusive sleep correlates monitoring", submitted to IEEE Sensors Journal

10. Marinai, C. and Melissa, E., Arcarisi, L., Bossi, F., Bufano, P., Di Rienzo, F., Zanoletti, M., Greco, A., Vallati, C., Carbonaro, N., Laurino, M., & Tognetti, A. (2025). In-bed unobtrusive heart rate estimation using ballistocardiography and sensor fusion algorithm. Accepted to "E-Textiles: International Conference on the Challenges, Opportunities, Innovations and Applications in Electronic Textiles" (2025)

For the most current list of publications, please visit the project website: https://www.tolife-project.eu/

## Usage Guidelines

### Recommended Use Cases
- Longitudinal health monitoring research
- Sleep pattern analysis
- Environmental health impact studies
- Digital health technology validation
- Machine learning model development for health prediction and exacerbation detection

### Citation
If you use this dataset in your research, please cite:

```
TOLIFE Consortium. (2026). Health Indices Dataset from Clinical Study A - TOLIFE Project (Version 2.0).
Continuous monitoring of health parameters for COPD patients using unobtrusive sensors.
Zenodo. https://doi.org/10.5281/zenodo.23263785
```

The DOI above identifies version 2.0. To refer to the dataset regardless of version, use https://doi.org/10.5281/zenodo.16642438, which always resolves to the latest version.

### License
This dataset is released under Creative Commons Attribution 4.0 International (CC BY 4.0) license.

## Data Availability Statement

This dataset is publicly available through Zenodo. Version 2.0: https://doi.org/10.5281/zenodo.23263785. Latest version: https://doi.org/10.5281/zenodo.16642438.

## Contact Information

**Principal Investigator**: TOLIFE Consortium
**Data Contact**: For data-related inquiries, please contact the TOLIFE project team
**Institution**: University of Pisa, Italy (Lead Institution) and participating European institutions
**Project Website**: https://www.tolife-project.eu/

## Acknowledgments

The work is supported by European Union's Horizon Europe Research and Innovation Programme under grant agreement No. 101057103 – project TOLIFE.

We acknowledge the valuable contribution of all participant volunteers and the technical and clinical staff involved in data collection across Germany, Spain and Italy.

## Version History

- **v2.0** (2026): Complete Clinical Study A release
  - Monitoring period extended to the full study (May 2024 - April 2026)
  - 86 participants (12 more than v1.0), including the Italian site
  - Sensor indices regenerated for the whole study with the final version of the processing pipeline
  - New daily estimated clinical scores: `CAT.csv`, `CCQ.csv`, `mMRC.csv`
  - Clinical visits (T1-T5) and exacerbation records updated to the end of the study
  - New dataset-level files: `kits.csv`, `clinical_indices_all.csv`, `exacerbations_all.csv`
  - Timestamp column name harmonised to `Timestamp` in all index files
  - Validity flags harmonised to 0/1 integers in all index files

- **v1.0** (July 2025): Initial dataset release
  - First year of monitoring, 74 participants (Germany and Spain)
  - Consolidated daily measurements into participant-level files
  - Standardized timestamps to UTC format
  - Quality flags included for all measurements

## Known Issues and Limitations

- Some participants have missing data periods due to participant non-compliance or device issues
- Data quality varies across participants and measurement periods (indicated by validity flags)
- Participants enrolled late in the study have shorter monitoring periods
- Environmental and mobility measurements may be affected by local conditions and participant lifestyle
- Clinical assessments are available only at specific timepoints (T1-T5) and not continuously

## Support and Feedback

For questions about this dataset or to report issues, please:
- Visit the project website: https://www.tolife-project.eu/
- Contact the TOLIFE project team through the official project channels

---

*Last updated: September 22, 2026*