The　classification　of　biomedical　information　plays　an　important　role　in　the　prediction　and　prevention　of　various　physiological　and　psychological　diseases.　SVM　is　widely　used　in　biomedical　information　classification　due　to　its　strong　practicability　in　solving　data　classification　problems　such　as　small　sample,　nonlinearity　and　high　dimension.　To　improve　the　classification　accuracy　of　SVM　in　biomedical　information,　a　particle　swarm　optimization　algorithm　based　on　multi-population　mutation　(MsM-PSO)　is　proposed　in　this　paper.　MsM-PSO　uses　multiple　subpopulations　to　search　the　optimal　solution　in　parallel.　When　nearly　half　of　the　subpopulations　are　clustered,　The　Gaussian　mutation　is　performed　on　the　optimal　particle　in　each　subpopulation,　while　the　feedback　mutations　are　performed　on　the　two　remaining　poorer　particles　in　each　subpopulation.　Then　the　improved　PSO　algorithm　is　used　to　optimize　the　parameters　of　the　SVM　model.　A　new　classification　method　(MsM-PSO-SVM)　is　proposed.　To　verify　the　classification　performance　of　the　MsM-PSO-SVM,　this　article　classifies　biomedical　data.　The　test　result　shows　that　the　proposed　MsM-PSO-SVM　has　achieved　satisfactory　classification　result　in　biomedical　prediction.