2 posts tagged with "web API"

We are happy to announce that we have published version three of Cozie-Apple. We worked towards a cleaner, simpler code base in the app and the backend while also improving reliability. We have already started working with Cozie v3 and it is already deployed for research.

New features in Cozie v3​

• New HealthKit data points
  • Heart rate variability (HRV)
  • Wrist temperature
  • Sleep analysis
• Offline data logging (in-app file download)
• Customizable backend configuration in the app
• Customizable watch surveys using JSON files
• Customizable phone survey using browser-based surveys (e.g., Qualtrics, Google Forms, etc.)

We have added more content to the website:

• Deployment guide
• Frequently asked questions (FAQ)
• Troubleshooting guide

We have updated the web APi for Cozie data retrieval. The change became necessary as the requests from some users exceeded the current payload limit of 6 MB. The example script was changed in to places:

1. The 'shutil' module needs to be imported at the beginning

import shutil

2. The new web API doesn't serve the data as JSON string. Instead, it is a two-step process. Firstly, a link is provided to a zipped .csv-file containing the data. Hence, the provided link needs to be called to then download the actual data.

# Download zipped CSV file with Cozie data
with requests.get(url, stream=True) as r:
    with open('cozie.zip', 'wb') as f:
        shutil.copyfileobj(r.raw, f)

# Convert zipped CSV file with Cozie to dataframe
with open('cozie.zip', 'rb') as f:
      df = pd.read_csv(f, compression={'method': 'zip', 'archive_name': 'sample.csv'})

df = df.drop(columns=['Unnamed: 0'])

The initial call of the web API and the timezone manipulation of the dataframe remains the same, as does the resulting Pandas dataframe.