Quantifying the thermal-ventilation trade-off and overnight CO 2 exposure in Korean jjokbang substandard housing

The thermal-ventilation trade-off has rarely been quantified at event level, and overnight CO 2 exposure in vulnerable elderly residents of substandard housing is under-monitored. We monitored 27 Korean jjokbang households ( n = 25 valid; male medical-aid recipients aged 65+) in Daegu over 91 winter days using 5-min Internet of Things sensors. Candidate 5-minute records (CO 2 peak > 1500 ppm and drop > 300 ppm) were merged into 4693 distinct ventilation episodes (median duration 30 min). Per episode, the median CO 2 reduction was 910 ppm at a median temperature cost of 0.51 °C, with 5th-percentile cooling of −2.52 °C; 7.3% of episodes transitioned from above to below 18 °C and 27.0% remained below it throughout. A household-level cluster bootstrap gave a marginal association of −218 ppm °C −1 (95% CI −357 to −151), explaining 9.8% of variance. Sleep-window CO 2 was invariant across four definitions (per-household median 1284–1379 ppm; 95th percentile 2363–2443 ppm). Within 22:00–06:00, 5/25 (20%) stayed below the 1000 ppm bedroom reference level and 8/25 (32%) exceeded 2000 ppm. These results place this population at the upper end of published bedroom-CO 2 ranges and provide reference points for heat-recovery and bedroom-ventilation interventions in cold-climate substandard housing.