Objective To explore the application value of combined speech and electrocardiogram (ECG) feature analysis in the early screening of mild cognitive impairment (MCI) among older adults, and to provide a reference for early diagnosis and individualized nursing interventions of MCI. Methods A convenience sampling method was adopted to recruit 365 community-dwelling older adults from Qixia District, Nanjing, between July and September 2024. Based on the Montreal Cognitive Assessment-Beijing (MoCA-B), participants were classified into an MCI group (n=253) and a cognitively normal group (n=112). Speech data were collected through the "Cookie Theft Picture" task, and ECG signals were synchronously recorded. Multivariate logistic regression analysis was used to construct working characteristic curves and calculate the area under the ROC curve (AUC), to evaluate the predictive value of speech and ECG features for MCI in older adults. Results Multivariate logistic regression analysis indicated that, long pauses/total speech duration, loudness_sma3_percentile50.0, matching score, HRV_IQRNN, and HRV_HTI were major influencing factors of MCI (all P<0.05). ROC curve analysis showed that, the AUC for speech features and ECG features were 0.694 and 0.625, respectively, and the combined screening performance was highest with an AUC of 0.728. Conclusion The combined analysis of speech and ECG features can effectively improve the screening accuracy for MCI in older adults, so it provides a new perspective for early screening of high-risk populations.