**5. Conclusions**

While CWT based gait analysis methods have been widely adopted in previous studies, their performances in gait event detection when using different mother wavelets have rarely been studied, especially in hemiplegic patients. In this study, different mother wavelets and wavelet selection criteria were systematically investigated by using acceleration signals recorded from both hemiplegic and healthy subjects. The experimental results demonstrated that an overall significant difference in performance of the CWT algorithm was observed when using different mother wavelet functions for detecting HS and TO gait events, which suggested the need for this study. Additionally, we found that the accuracy criteria based on time-error minimization and F1-score maximization led to the realization of an appropriate mother wavelet named as "db6" which achieved the highest detection accuracy with lowest detection time-error for both hemiplegic and healthy subjects. The outcomes of this study may provide an insight on mother wavelet selection criteria for gait event analysis especially in hemiplegic patients, and may ultimately facilitate the practical development of rehabilitation devices or strategies for them.

**Author Contributions:** Conceptualization, H.Z., N.J., Z.H., and G.L.; investigation, N.J., H.Z., K.G., L.X., and G.L.; methodology, N.J. and H.Z.; software, N.J.; validation, N.J.; patients' recruitment, K.G. and Z.H.; data collection, N.J., K.G, and Z.H.; data analysis, N.J., H.Z., O.W.S., and Z.H.; data interpretation, K.G. and G.L.; writing—original draft preparation, N.J.; writing—review and editing, G.L., L.X., N.J., and O.W.S.; supervision, G.L. and L.X.; funding acquisition, G.L. and Z.H.

**Funding:** The work was supported in part by the National Natural Science Foundation of China under Grant (#U1613222, #61773110), Shenzhen Science and Technology Plan Project (#JCYJ20160331174854880, #JCYJ20160331185848286), Science and Technology Program of Guangzhou (#201803010093), and Fundamental Research Funds for the Central Universities Grant (#N181906001).

**Acknowledgments:** The authors would thank the staffs at the CAS Key Lab of Human-Machine Intelligence-Synergy Systems of Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences (CAS), and Guangzhou Panyu Central Hospital for their support in data acquisition.

**Conflicts of Interest:** The authors declare no conflict of interest.
