Improving behavioral assessment in genetically modified mice.

通过维数减少统计分析，我已经确定，到目前为止，老鼠的运动活动和类似焦虑的行为在很大程度上是两种无法区分的结构。这一发现非常有趣。简而言之，这一结果表明，几乎所有与精神疾病小鼠模型中的焦虑样行为有关的结论，至少在很大程度上是没有意义的。因此，为了提高小鼠行为研究的可信度，未来的研究将是必要的。具体来说，在小鼠模型研究中应该开发和引入一些与小鼠焦虑非运动活动相关的测量方法，以便可靠地分离上述两种构念。第二个研究成果是关于如何在时间序列设置中正确建模小鼠运动活动。该领域的大多数研究人员采用简单的方差分析模型。遗憾的是，这些模型存在严重的缺陷，包括不现实的统计假设（例如线性和高斯分布）。我设计了一个广义加性模型定位形状和尺度（GAMLSS）模型，显著提高了结果的准确性。这些结果已在若干国际会议上传播。

I have established, via dimensionality reduction statistical analyses, that locomotor activity and anxiety-like behaviour in mice are, to date, two indistinguishable constructs to a large extent. This finding is of extreme interest. In a nutshell, this outcome leads to the ascertainment that pretty much all conclusions relative to anxiety-like behaviour in mice models of psychiatric disease are, at least to a large extent, moot. Consequently, future research will be necessary in order to improve the trustworthiness of mouse behaviour research. Specifically, some non-locomotor-activity-related measurements of mouse anxiety should be developed and introduced in mouse model research in order to reliably dissociate between the two above constructs.The second research achievement refers to how correctly model mouse locomotor activity in time series settings. Most researchers in the field employ simple ANOVA models. Regrettably, such models suffer from serious shortcomings including unrealistic statistical assumptions (e.g., linearity and Gaussian distribution). I have designed a Generalized Additive Model Location Shape and Scale (GAMLSS) model that significantly improve the accuracy of the results.These results have been disseminated in several international conferences.