Recently, the adoption of micromobility as an alternative mode of transportation on a large scale has been growing rapidly. However, its operational and safety aspects have not been extensively investigated in the literature. Following this purpose, we developed a novel methodology that aims at evaluating priority areas for shared micromobility system users’ accident risk mitigation based on predicted injury severity using a machine learning-based approach. The methodology proposed in this paper consists of two models: a predictive model, which is based on an artificial neural network with a pattern recognition algorithm, to estimate the expected safety indicator of an urban zone, and a clustering method to define the priority areas for intervention through the application of a geofence speed regulation system. A real case study was carried out in the city of Bari, Italy, to test the effectiveness of the proposed methodology. The results showed how it is possible to define areas for intervention with different priorities based on the expected severity index. The proposed methodology can be seen as a decision support system to assist transport operators and urban planners in regulating shared micromobility vehicles in urban areas by defining priority areas for intervention through geofencing and, therefore, it can be useful for improving micromobility adoption, road safety, and urban mobility policies.

中文翻译：

---

一种基于伤害严重程度的方法，用于评估共享微出行事故风险缓解的优先领域


最近，微出行作为一种替代交通方式的大规模采用正在迅速增长。然而，其操作和安全方面尚未在文献中得到广泛研究。为此，我们开发了一种新方法，旨在使用基于机器学习的方法，根据预测的伤害严重程度来评估共享微出行系统用户事故风险缓解的优先领域。本文提出的方法由两个模型组成：一个预测模型，它基于带有模式识别算法的人工神经网络，用于估计城市区域的预期安全指标，以及一种聚类方法，通过应用地理围栏速度调节系统来定义优先干预区域。在意大利巴里市进行了一项真实的案例研究，以测试所提出的方法的有效性。结果表明，如何根据预期的严重性指数定义具有不同优先级的干预区域。所提出的方法可以被视为一种决策支持系统，通过地理围栏定义干预的优先区域，帮助交通运营商和城市规划者监管城市地区的共享微出行车辆，因此，它可用于改善微出行的采用、道路安全和城市交通政策。