Metabolite analysis of body fluids is an advanced method for disease diagnosis and status assessment. Laser desorption/ionization-mass spectrometry (LDI-MS) has been widely employed for metabolic analysis due to the fast detection speed and simple sample pretreatment. Here, we designed and synthesized hollow mesoporous carbon nanospheres anchored with Ni (HMCSs/Ni) to simultaneously enhance the ionization and thermal desorption processes of the LDI process owing to their hollow and mesoporous structure, large surface area, and abundant Ni-N bonds. Based on HMCSs/Ni, we built an LDI-MS platform that can be used for metabolic information extraction and achieved the rapid detection (about seconds per sample) of metabolic fingerprints in trace serum samples (∼0.1 μL) without complicated preprocessing procedures. Then, we conducted serum metabolic screening in a cohort of COVID-19-recovered pregnant women. The optimized machine learning model could distinguish recovered pregnant women from uninfected pregnant women based on metabolic features with an AUC value of 0.901. In addition, the model indicates that maternal COVID-19 infection does not significantly affect the metabolic fingerprints of the fetuses. Overall, our work shows the prospect of HMCSs/Ni-assisted LDI-MS in disease recovery assessment and metabolite analysis.