基于生存分析的烤烟上部叶褐变启动时间模型研究

    A survival analysis-based model for browning onset time in flue-cured tobacco upper leaves

    • 摘要:
      目的 研究烤烟上部叶褐变启动时间分布特征,确定关键影响因素与褐变启动时间的关系。
      方法 采用双因素随机组合设计,设置不同成熟度(欠熟、成熟、过熟)和定色期干球温度(40、42、44 ℃),连续监测烘烤过程中烟叶图像变化,以叶片出现褐色斑点作为褐变启动判据,采用Kaplan-Meier法对褐变启动时间进行单因素分析,并建立基于Cox回归分析的褐变启动时间模型。
      结果 成熟度对褐变启动风险无显著影响,与欠熟叶相比,成熟叶和过熟叶的风险比分别为0.85和1.10,且均未达到显著差异。定色期干球温度是影响‘云烟87’上部叶褐变启动的关键风险因素,44 ℃条件下褐变发生比例最高,生存曲线下降速度最快,且与40 ℃相比,44 ℃使褐变启动风险增加约1.52倍,而42 ℃与40 ℃无显著差异。模型诊断结果表明,各变量及全局比例风险假设检验P值均大于0.05,满足Cox模型的基本假设,参数估计结果可靠。
      结论 本研究给出了褐变启动时间风险函数与影响因素之间的定量关系,为上部叶烘烤参数优化提供了技术支撑和理论参考。

       

      Abstract:
      Objective To investigate the distribution characteristics of browning onset time in the upper leaves of flue-cured tobacco, and determine the relationships between key influencing factors and browning onset time.
      Method A two-factor randomized combination design was employed, with different maturity levels (unripe, mature, overripe) and dry-bulb temperatures at the color-fixing stage (40, 42, 44 ℃). Image changes of tobacco leaves were continuously monitored during the curing process, with the appearance of brown spots on the leaf surface as the criterion for browning onset. The Kaplan-Meier method was employed for univariate analysis of browning onset time, and a model for browning onset time was established based on Cox regression analysis.
      Result Maturity level had no significant effect on the risk of browning onset . Compared with unripe leaves, the hazard ratios for mature and overripe leaves were 0.85 and 1.10, respectively, and neither was statistically significant. The dry-bulb temperature at the color-fixing stage was a key risk factor affecting browning onset in the upper leaves of ‘Yunyan 87’. At 44 ℃, the proportion of browning occurrence was the highest, and the survival curve declined fastest. Compared with 40 ℃, 44 ℃ increased the risk of browning onset by approximately 1.52 times, while the difference between 42 ℃ and 40 ℃ was not significant. The model diagnostics showed that the P values for both individual variables and the global proportional hazards assumption test were greater than 0.05, indicating that the proportional hazards assumption was not violated and the parameter estimates of the Cox model were reliable.
      Conclusion This study provides a quantitative relationship between the risk function of browning onset time and the influencing factors, providing technical support and a theoretical reference for optimizing the curing parameters for upper leaves.

       

    /

    返回文章
    返回