forecose

A time-series forecasting extension for pydexcom using Google's TimesFM. Readings from the previous 24 hours are captured from the Dexcom Share API service and fed into the model to create a forecast of blood glucose values for the next hour.

All modelling and forecasting is performed locally on your device. The only external connections made are with:

• Dexcom Share API: fetching CGM readings following the pydexcom approach.
• HuggingFace: one-time download of the forecasting model weights on the first run.

Quick Start

1. Ensure that you have also installed the pydexcom package and enabled the Share service within your Dexcom G7 / G6 / G5 / G4 mobile app.

pip install pydexcom forecose

2. Initialise pydexcom with your Dexcom credentials (below shows the simplist route, refer to pydexcom for further instruction).

>>> from pydexcom import Dexcom
>>> dexcom = Dexcom(username="username", password="password")

3. Generate a prediction. By default, the DexcomForecast class predicts the upcoming hour (the next 12 readings) to prevent overextending the forecast. Set custom prediction lengths by adjusting the horizon argument.

>>> from forecose import DexcomForecast
>>> predictions = DexcomForecast(horizon=12).get_forecast(dexcom)
>>> print(predictions)
                          timestamp  predicted_glucose    q10    q20    q30    q40    q50    q60    q70    q80    q90
0  2026-10-01 09:07:45.945000+01:00              161.0  158.0  160.0  160.0  160.0  161.0  162.0  163.0  166.0  171.0
1  2026-10-01 09:12:45.894000+01:00              163.0  154.0  158.0  159.0  161.0  163.0  165.0  168.0  171.0  178.0
2  2026-10-01 09:17:45.843000+01:00              164.0  151.0  156.0  159.0  161.0  164.0  167.0  170.0  175.0  184.0
3  2026-10-01 09:22:45.792000+01:00              165.0  148.0  155.0  159.0  162.0  165.0  169.0  173.0  179.0  190.0
4  2026-10-01 09:27:45.741000+01:00              167.0  146.0  153.0  158.0  162.0  167.0  171.0  176.0  183.0  194.0
5  2026-10-01 09:32:45.690000+01:00              167.0  143.0  152.0  158.0  162.0  167.0  172.0  177.0  185.0  199.0
6  2026-10-01 09:37:45.639000+01:00              167.0  140.0  149.0  155.0  161.0  167.0  172.0  179.0  187.0  201.0
7  2026-10-01 09:42:45.588000+01:00              167.0  138.0  148.0  155.0  161.0  167.0  172.0  179.0  189.0  204.0
8  2026-10-01 09:47:45.537000+01:00              167.0  136.0  147.0  154.0  160.0  167.0  173.0  181.0  190.0  207.0
9  2026-10-01 09:52:45.486000+01:00              167.0  133.0  146.0  152.0  160.0  167.0  174.0  182.0  191.0  209.0
10 2026-10-01 09:57:45.435000+01:00              167.0  132.0  145.0  152.0  160.0  167.0  175.0  184.0  193.0  212.0
11 2026-10-01 10:02:45.384000+01:00              166.0  129.0  143.0  150.0  158.0  166.0  175.0  184.0  194.0  214.0

>>> print(predictions.mmol_l)
                          timestamp  predicted_glucose  q10  q20  q30  q40  q50  q60   q70   q80   q90
0  2026-10-01 09:07:45.945000+01:00                8.9  8.8  8.9  8.9  8.9  8.9  9.0   9.0   9.2   9.5
1  2026-10-01 09:12:45.894000+01:00                9.0  8.5  8.8  8.8  8.9  9.0  9.2   9.3   9.5   9.9
2  2026-10-01 09:17:45.843000+01:00                9.1  8.4  8.7  8.8  8.9  9.1  9.3   9.4   9.7  10.2
3  2026-10-01 09:22:45.792000+01:00                9.2  8.2  8.6  8.8  9.0  9.2  9.4   9.6   9.9  10.5
4  2026-10-01 09:27:45.741000+01:00                9.3  8.1  8.5  8.8  9.0  9.3  9.5   9.8  10.2  10.8
5  2026-10-01 09:32:45.690000+01:00                9.3  7.9  8.4  8.8  9.0  9.3  9.5   9.8  10.3  11.0
6  2026-10-01 09:37:45.639000+01:00                9.3  7.8  8.3  8.6  8.9  9.3  9.5   9.9  10.4  11.2
7  2026-10-01 09:42:45.588000+01:00                9.3  7.7  8.2  8.6  8.9  9.3  9.5   9.9  10.5  11.3
8  2026-10-01 09:47:45.537000+01:00                9.3  7.5  8.2  8.5  8.9  9.3  9.6  10.0  10.5  11.5
9  2026-10-01 09:52:45.486000+01:00                9.3  7.4  8.1  8.4  8.9  9.3  9.7  10.1  10.6  11.6
10 2026-10-01 09:57:45.435000+01:00                9.3  7.3  8.0  8.4  8.9  9.3  9.7  10.2  10.7  11.8
11 2026-10-01 10:02:45.384000+01:00                9.2  7.2  7.9  8.3  8.8  9.2  9.7  10.2  10.8  11.9

What do these predictions mean?

• predicted-glucose: The average trajectory of your blood sugar forecast (smoothed, centred baseline of the confidence bands).
• q10 to q90: The range of confidence bands provide a realistic upper and lower estimate boundaries, showing the full probability distribution of predicted glucose values.

Event Modelling

To account for key events (e.g., insulin administration or carbohydrate (carbs) intake) that act on blood glucose values without distorting the underlying TimesFM probability distribution, you can apply a deterministic overlay to your baseline forecast.

Drawing on mathematical frameworks utilised in closed-loop Artifical Pancreas systems and the Hovorka/Bergman meal submodels, event impacts are computed as a second-order linear delay process. Here, event unit rates (e.g., the absorption of insulin or carbs) are translated into a physiological curve that begins slowly, reaches a peak, and then gradually decays over time.

By default, forecose updates the forecast predictions using standard clinical baselines (a 55-minute peak for insulin, and a 40-minute peak for carbs):

>>> carb_predictions = predictions.add_event(type="carbs", units=30, minutes_ago=0)
>>> print(carb_predictions.mmol_l)
                          timestamp  predicted_glucose   q10   q20   q30   q40   q50   q60   q70   q80   q90
0  2026-10-01 09:07:45.945000+01:00                9.0   8.8   8.9   8.9   8.9   9.0   9.0   9.1   9.3   9.5
1  2026-10-01 09:12:45.894000+01:00                9.2   8.7   8.9   9.0   9.1   9.2   9.3   9.5   9.7  10.0
2  2026-10-01 09:17:45.843000+01:00                9.5   8.8   9.0   9.2   9.3   9.5   9.7   9.8  10.1  10.6
3  2026-10-01 09:22:45.792000+01:00                9.8   8.8   9.2   9.4   9.6   9.8  10.0  10.2  10.5  11.2
4  2026-10-01 09:27:45.741000+01:00               10.2   9.0   9.4   9.7   9.9  10.2  10.4  10.7  11.0  11.7
5  2026-10-01 09:32:45.690000+01:00               10.4   9.1   9.6   9.9  10.2  10.4  10.7  11.0  11.4  12.2
6  2026-10-01 09:37:45.639000+01:00               10.7   9.2   9.7  10.0  10.4  10.7  11.0  11.4  11.8  12.6
7  2026-10-01 09:42:45.588000+01:00               11.0   9.4  10.0  10.4  10.7  11.0  11.3  11.7  12.3  13.1
8  2026-10-01 09:47:45.537000+01:00               11.3   9.6  10.2  10.6  10.9  11.3  11.7  12.1  12.6  13.5
9  2026-10-01 09:52:45.486000+01:00               11.7   9.8  10.5  10.8  11.3  11.7  12.0  12.5  13.0  14.0
10 2026-10-01 09:57:45.435000+01:00               11.9  10.0  10.7  11.1  11.5  11.9  12.4  12.9  13.4  14.4
11 2026-10-01 10:02:45.384000+01:00               12.2  10.1  10.9  11.3  11.7  12.2  12.7  13.2  13.7  14.8

>>> insulin_predictions = predictions.add_event(type="insulin", units=5, minutes_ago=0)
>>> print(insulin_predictions.mmol_l)
                          timestamp  predicted_glucose  q10  q20  q30  q40  q50  q60  q70  q80   q90
0  2026-10-01 09:07:45.945000+01:00                8.9  8.7  8.8  8.8  8.8  8.9  8.9  9.0  9.2   9.4
1  2026-10-01 09:12:45.894000+01:00                8.9  8.4  8.6  8.7  8.8  8.9  9.0  9.2  9.3   9.7
2  2026-10-01 09:17:45.843000+01:00                8.8  8.0  8.3  8.5  8.6  8.8  8.9  9.1  9.4   9.9
3  2026-10-01 09:22:45.792000+01:00                8.6  7.7  8.0  8.3  8.4  8.6  8.8  9.0  9.4  10.0
4  2026-10-01 09:27:45.741000+01:00                8.4  7.3  7.7  7.9  8.2  8.4  8.7  8.9  9.3   9.9
5  2026-10-01 09:32:45.690000+01:00                8.1  6.8  7.3  7.6  7.8  8.1  8.4  8.7  9.1   9.9
6  2026-10-01 09:37:45.639000+01:00                7.8  6.3  6.8  7.1  7.4  7.8  8.0  8.4  8.9   9.7
7  2026-10-01 09:42:45.588000+01:00                7.4  5.8  6.4  6.8  7.1  7.4  7.7  8.1  8.7   9.5
8  2026-10-01 09:47:45.537000+01:00                7.0  5.3  5.9  6.3  6.7  7.0  7.4  7.8  8.3   9.3
9  2026-10-01 09:52:45.486000+01:00                6.7  4.8  5.6  5.9  6.3  6.7  7.1  7.5  8.0   9.0
10 2026-10-01 09:57:45.435000+01:00                6.3  4.4  5.1  5.5  5.9  6.3  6.8  7.3  7.8   8.8
11 2026-10-01 10:02:45.384000+01:00                5.9  3.8  4.6  5.0  5.4  5.9  6.4  6.9  7.4   8.5

Sensitivity to insulin (ISF) and carbohydrates (CSF) is set at 40 mg/dL per 1U and 4 mg/dL per gram, respectively. These values are placeholders meant to represent reasonable values and should be adjusted through the add_event method using the tau and sensitivity parameters. In future, I hope to develop a method for calculating estimate values from historic data.

>>> insensitive_insulin_predictions = predictions.add_event(type="insulin", units=5, minutes_ago=0, sensitivity=20.0)
>>> print(insensitive_insulin_predictions.mmol_l)
                          timestamp  predicted_glucose  q10  q20  q30  q40  q50  q60  q70  q80   q90
0  2026-10-01 09:07:45.945000+01:00                8.9  8.8  8.9  8.9  8.9  8.9  9.0  9.0  9.2   9.5
1  2026-10-01 09:12:45.894000+01:00                9.0  8.5  8.7  8.8  8.9  9.0  9.1  9.3  9.4   9.8
2  2026-10-01 09:17:45.843000+01:00                8.9  8.2  8.5  8.7  8.8  8.9  9.1  9.3  9.5  10.0
3  2026-10-01 09:22:45.792000+01:00                8.9  7.9  8.3  8.5  8.7  8.9  9.1  9.3  9.7  10.3
4  2026-10-01 09:27:45.741000+01:00                8.8  7.7  8.0  8.3  8.5  8.8  9.0  9.3  9.7  10.3
5  2026-10-01 09:32:45.690000+01:00                8.7  7.4  7.9  8.2  8.4  8.7  9.0  9.3  9.7  10.5
6  2026-10-01 09:37:45.639000+01:00                8.5  7.0  7.5  7.9  8.2  8.5  8.8  9.2  9.7  10.4
7  2026-10-01 09:42:45.588000+01:00                8.3  6.7  7.3  7.7  8.0  8.3  8.6  9.0  9.5  10.4
8  2026-10-01 09:47:45.537000+01:00                8.2  6.4  7.0  7.4  7.8  8.2  8.5  8.9  9.4  10.4
9  2026-10-01 09:52:45.486000+01:00                8.0  6.1  6.8  7.2  7.6  8.0  8.4  8.8  9.3  10.3
10 2026-10-01 09:57:45.435000+01:00                7.8  5.9  6.6  7.0  7.4  7.8  8.3  8.8  9.3  10.3
11 2026-10-01 10:02:45.384000+01:00                7.5  5.5  6.3  6.7  7.1  7.5  8.0  8.5  9.1  10.2

Custom Data Forecasting

Forecast historical CGM data without the need to connect to the Dexcom Share API service. The input data must contain only Time and Glucose columns to successfully generate a forecast prediction.

>>> import pandas as pd
>>> custom_data = pd.read_csv("historical_cgm_data.csv")
>>> print(custom_data)
                     Time  Glucose
0     2026-08-27T00:01:57      6.7
1     2026-08-27T00:06:57      6.4
2     2026-08-27T00:11:57      6.2
3     2026-08-27T00:16:57      5.9
4     2026-08-27T00:21:58      5.8
...                   ...      ...
1714  2026-09-01T23:37:11      8.5
1715  2026-09-01T23:42:11      8.3
1716  2026-09-01T23:47:11      7.8
1717  2026-09-01T23:52:11      7.5
1718  2026-09-01T23:57:11      7.1

Pass the input data when initialising the DexcomForecast class via cgm_history to bypass the requirement to connect to the Dexcom Share API service.

>>> custom_predictions = DexcomForecast(cgm_history=custom_data).get_forecast()