As part of the George B. Moody PhysioNet Challenge 2026, our team, insomniac111, developed two open-source machine-learning classifiers to predict future cognitive impairment from polysomnography (PSG). The training data included a 1,103-record small data set and a 6,600-record large set from three institutions, with heterogeneous physiological signals, Complete AI Sleep Report (CAISR) annotations, human annotations, demographics, and diagnostic labels. Data preprocessing includes channel harmonization and 200-Hz resampling of EEG, EOG, and ECG. The first classifier, XGBoost, used a compact 156-feature representation: 12 demographic variables, 105 physiological summaries, 31 CAISR measures, and eight missingness indicators. The second classifier, TabPFN V2, used an expanded, quantile-normalized table of 243 features that additionally included regional EEG spectra, heart-rate variability, enhanced oxygenation, human annotations, and stage-conditional physiology. We used leave-one-site-out validation to estimate cross-source generalization. On the blinded evaluation, small-set trained XGBoost achieved an age-conditioned AUROC of 0.765 (Challenge Score), large-set XGBoost achieved 0.716, and small-set TabPFN V2 achieved 0.645. Notably, large-set XGBoost slightly improved conventional AUROC over small-set XGBoost (0.855 versus 0.851) while reducing the official challenge score (age-conditioned AUROC). These findings suggest that XGBoost and pre-trained transformer-based TabPFN, combined with multimodal PSG feature engineering, offer promising and computationally efficient approaches to early cognitive-impairment prediction from sleep studies.