A machine-learning photometric classifier for massive stars in nearby galaxies II. The catalog

Mass loss is a key aspect of stellar evolution, particularly in evolved massive stars, yet episodic mass loss remains poorly understood. To investigate this, we need evolved massive stellar populations across various galactic environments. However, spectral classifications are challenging to obtain in large numbers, especially for distant galaxies. We addressed this by leveraging machine-learning techniques. We combined Spitzer photometry and Pan-STARRS1 optical data to classify point sources in 26 galaxies within 5 Mpc, and a metallicity range 0.07-1.36 Z$\odot$. Gaia data release 3 (DR3) astrometry was used to remove foreground sources. Classifications are derived using a machine-learning model developed in our previous work. We report classifications for 1,147,650 sources, with 276,657 sources (~24%) being robust. Among these are 120,479 red supergiants (RSGs; ~11%). The classifier performs well even at low metallicities (~0.1 Z$\odot$) and distances under 1.5 Mpc, with a slight decrease in accuracy beyond ~3 Mpc due to Spitzer’s resolution limits. We also identified 21 luminous RSGs (log($L/L_\odot)\ge5.5$), 159 dusty yellow hypergiants in M31 and M33, as well as 6 extreme RSGs (log($L/L_\odot)\ge6$) in M31, challenging observed luminosity limits. Class trends with metallicity align with expectations, although biases exist. This catalog serves as a valuable resource for individual-object studies and James Webb Space Telescope target selection. It enables the follow-up on luminous RSGs and yellow hypergiants to refine our understanding of their evolutionary pathways. Additionally, we provide the largest spectroscopically confirmed catalog of extragalactic massive stars and candidates to date, beyond the Clouds, comprising 5,273 sources (including ~330 other objects).

💡 Research Summary

This paper presents the second installment of the ASSESS (Episodic Mass Loss in Evolved Massive stars: Key to Understanding the Explosive Early Universe) project, delivering a machine‑learning based photometric catalog of massive stars in 26 nearby galaxies within 5 Mpc. The authors combine mid‑infrared photometry from the Spitzer Space Telescope (IRAC 3.6, 4.5, 5.8, 8.0 µm and MIPS 24 µm) with optical data from Pan‑STARRS1 (g, r, i, z, y). To eliminate foreground Milky Way contaminants they exploit Gaia DR3 astrometry, fitting the distributions of parallax and proper motion for sources outside each galaxy’s visual ellipse, scaling this “foreground model” to the interior region, and flagging as foreground any source lying beyond three standard deviations of the Gaussian component that represents the galaxy’s own stellar population. For galaxies lacking sufficient Gaia detections they adopt the cuts derived for M31, the most densely sampled system.

The classification engine is the multi‑class random‑forest model introduced in Paper I. It was trained on spectroscopically confirmed massive stars in M31 and M33 (red supergiants, yellow supergiants, luminous blue variables, Wolf‑Rayet stars, etc.) together with non‑stellar contaminants, using a feature set that includes a suite of color indices (e.g.,