Meta-analysis of Frequency Learning Networks combined with co-depth point localization identification to screen and validate efficient preservation methods for pitaya: 1-methylcyclopropene, carvacrol and gibberellic acid

Abstract

This study verified the feasibility of Meta-analysis combined with co-depth point localization identification in screening plant preservation methods and innovative joint-use preservation strategies, pointed out the reasons for the deviation, optimized the computational method of network Meta-analysis, and explored the effectiveness of the combined treatments screened under the combination of the two approaches in practical application. A Bayesian random-effects model with Meta-analysis of Frequency Learning Networks (FLN) was used to compare the freshness preservation effects of pitaya under 28 different interventions, and the best interventions were screened for a single quality index of pitaya during the storage period. We combined the two treatments with the highest weights after the comparison and found, both experimentally and after validation with co-depth point localization identification, that the 1-methylcyclopropene (1-MCP) + carvacrol (CVR) coupled with gibberellic acid (GA₃) soaking was more effective in suppressing postharvest pitaya decay and maintaining quality during storage than 1-MCP + CVR treatment alone or soaking with GA₃.