Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

TimesFM

AI overview

Sign in and the AI will write an overview from our coverage.

Headlines · 3

  1. SensorWM supports personalized health coaching from wearable data

    Researchers developed SensorWM, an action-conditioned world model pre-trained on 2.28 million de-identified person-weeks of data from roughly 100,000 consenting Fitbit users. The team says that, given two weeks of baseline wearable data, it forecasts 20 physiological markers over 12 weeks under different daily steps, sleep, and cardio routines. The model outperforms general-purpose time-series foundation models such as TimesFM while preserving clinical patterns associated with diabetes, hypertension, and insulin resistance. SensorWM was paired with a feasibility-constrained optimizer and a language-model coaching agent to simulate 125 candidate habit combinations, filter them for realism and clinical safety, and produce progressive 12-week coaching plans.

    Google Research Blog · 🔥 0
  2. Google Earth AI expands early access with three experimental capabilities

    In its "Research@ 2026" post, the Google Research Blog said it is expanding early access to Google Earth AI and inviting partners to try three new experimental capabilities: deeper geospatial analysis inside Google Earth, risk assessments and scenario forecasting through a "planetary prediction engine," and an experimental agent-to-agent API that puts Geospatial Reasoning agent skills such as satellite imagery parsing, routing and demographic embeddings into custom workflows. Google says these run on more than 70 geospatial tools, including foundation models RSFM (Remote Sensing Foundation Model) and PDFM (Population Dynamics Foundation Model), alongside Google Dataset Search for querying more than 100 million public datasets and TimesFM for planetary nowcasting and forecasting. Google says the capabilities are validated by empirical benchmarks in its Earth AI Technical Report, and Google.org is providing $10 million for crisis resilience and public health work, including $2 million to The Milken Institute.

    Google Research Blog · 🔥 0
  3. Google Research details Earth AI geospatial models across public health studies

    A Google Research blog post describes how its Earth AI planetary geospatial foundation models were applied in a set of public health collaborations. In an MMR vaccination cross-border extrapolation study across 146 US–Canada border counties, capturing cross-border mobility and information spillovers produced a 36% relative gain in explained variance (0.159 to 0.216, statistically significant). Cardiovascular mortality nowcasting across 3,091 US counties was comparable to census-based models (mean absolute error 18.7 vs 19.1 deaths per county, not statistically significant), while dengue forecasting in Mexico gained a statistically significant one-month improvement through TimesFM integration, with accuracy improving in up to 72% of active transmission municipalities. Other cases include postpartum depression risk prediction with the University of Washington on 332,970 CDC PRAMS respondents (AUC +0.0020 and +0.0038) and cholera onset prediction with WHO AFRO across 403 health zones in the DRC (+9.7% AUPRC at four weeks).

    Google Research Blog · 🔥 0

Experience and discussion from the community

Share my TimesFM experienceAsk about TimesFM

timesfm: A pretrained time-series foundation model for time-series forecasting.

What problem it solves Time-series forecasting requires models that can forecast univariate and multivariate series across different context lengths. Forecasters also need to incorporate optional covariates and produce quantile forecasts without requiring per-task tuning. Key features Univariate and multivariate forecasting — Forecast individual time series or multiple target variates from NumPy a…

github.comhttps://github.com/google-research/timesfm