📓 Implementation Example

Let's look at a complete scenario with a user.

At first, we start by activating the device and we pass 80kg as the baseline weight.

Endpoint Details:

• Endpoint: https://wbsapi.withings.net/v2/user
• Method: POST

{
    "action":  "activate",
    "client_id":  "yourClientID",  
    "mailingpref":  1,  
    "birthdate":  "1985-06-15",
    "nonce": nonce,
    "signature": signature,
    "measures":  '[{"value":190,"unit":-2,"type":4},{"value":10,"unit":0,"type":1}]',  
    "gender":  0,  
    "preflang":  "en_US",  
    "unit_pref": '{"weight":1,"height":6,"distance":6,"temperature":11}',  
    "email":  "user@example.com",  
    "timezone":  "America/New_York",  
    "shortname":  "JDE",  
    "mac_addresses":  ["00:1A:2B:3C:4D:5E"],  
}

We get our authorization code, and we can exchange it for an access and refresh token.

Then, using our tokens, we can pull the measurement and check the probability:

{
    "status": 0,
    "body": {
        "updatetime": 1712722989,
        "timezone": "America/New_York",
        "measuregrps":  [
            {
                "grpid": 5423702714,
                "attrib": 0,
                "date": 1712722953,
                "created": 1712722989,
                "modified": 1712722989,
                "category": 1,
                "deviceid": 14546187,
                "hash_deviceid": "7df35c0a18c61a7009072621879658d91467dcb3",
                "probability": 100,
                "measures":  [
                    {
                        "value": 75232,
                        "type": 1,
                        "unit": -3,
                        "algo": 218236163,
                        "fm": 3
                    }
                ],
                "modelid": 17,
                "model": "Body Comp Pro",
                "comment": null
            }
        ]
    }
}

As expected, we get a probability of 100.

Now, lets say that the user step on the scale for the first time.

We will receive a notification, and we can pull the first "real measurement":

{
  "status": 0,
  "body": {
    "updatetime": 1712722989,
    "timezone": "America/New_York",
    "measuregrps": [
        {
            "grpid": 5423702714,
            "attrib": 0,
            "date": 1712722953,
            "created": 1712722989,
            "modified": 1712722989,
            "category": 1,
            "deviceid": 14546187,
            "hash_deviceid": "7df35c0a18c61a7009072621879658d91467dcb3",
            "probability": 100,
            "measures": [
            {
             "value": 75232,
             "type": 1,
             "unit": -3,
             "algo": 218236163,
             "fm": 3
            }],
        "modelid": 17,
        "model": "Body Comp Pro",
        "comment": null
    }]
    }
}

We have a probability of 92 for this measurement.

A good pratice is to always confirm the first measurement to ensure that the baseline weight is correct.

In order to do so we recommend you to prompt the user with a message in your application or to send him a push notification/sms.

Once the user confirmed the measurement on your side, we can perform the request to the Withings API:

{
    "access_token": "YourAccessToken",
    "action": "confirm",
    "grpid": "12345",
    "is_confirmed": 1
}

Now, we should get high probabilty for the next measurements if it's the same user using the scale.

Let's say that someone else use the scale to weight himself.

We will receive a notification and we can pull the measurement:

{
  "status": 0,
  "body": {
    "updatetime": 1712722989,
    "timezone": "America/New_York",
    "measuregrps": [
        {
            "grpid": 5423702714,
            "attrib": 0,
            "date": 1712722953,
            "created": 1712722989,
            "modified": 1712722989,
            "category": 1,
            "deviceid": 14546187,
            "hash_deviceid": "7df35c0a18c61a7009072621879658d91467dcb3",
            "probability": 23,
            "measures": [
            {
             "value": 75232,
             "type": 1,
             "unit": -3,
             "algo": 218236163,
             "fm": 3
            }],
        "modelid": 17,
        "model": "Body Comp Pro",
        "comment": null
    }]
    }
}

We get a low probability as expected.

Once we know this measurement doesn't belong to the user with can do the request to reject the measurement:

{
  "access_token": "YourAccessToken",
  "action": "confirm",
  "grpid": "12345",
  "is_confirmed": 0
}

This will improve the algorithm for future measurements.