55 Falls / Ambient Assisted Living

Background 55 Falls / Ambient Assisted Living examines the intersection of care, technology, and simulation in the age of machine learning. The work replays footage from a 2016 Belgian healthcare dataset—“High Quality Fall Simulation Data”—in which researchers at KU Leuven’s Advanced Integrated Sensing Lab perform staged falls to train fall-detection algorithms. Through this material, the project exposes the uncanny overlap of technical procedure and theatrical performance: scientists enacting imagined frailty for the machinic gaze. The work extends Machine Listening’s investigation into how datasets shape perception, knowledge, and power, asking what happens when gestures of care become data points and when bodies are rendered as predictive models for ageing, vulnerability, and mortality. Contribution Developed by Machine Listening (Sean Dockray, James Parker, Joel Stern), the two-channel video installation combines the original research footage with generative-AI re-creations of the same scenes. Musicians Lizzy Welsh (violin), Jessica Aszodi (voice), and Joel Stern (synthesiser) perform a live soundtrack for a film never intended for human viewing. The work work examines converging failures: the human inability to convincingly perform frailty and the machine’s inability to render it. These doubled inadequacies become a site for reflecting on embodiment, empathy, and the algorithmic mediation of care. SignificancePresented in The Mourning After (RMIT Design Hub Gallery, 24 Jul – 20 Sep 2025, curated by Larissa Hjorth), 55 Falls contributed to a collective exploration of grief, ritual, and loss in a world reshaped by technology. The work reimagines care as both human relationship and technological abstraction, revealing how, in the algorithmic age, intimacy, dependency, and mortality are reconfigured when life itself becomes training data.

背景：《55次跌倒/环境辅助生活（55 Falls / Ambient Assisted Living）》聚焦机器学习时代下护理、技术与仿真三者的交叉研究领域。该项目重放2016年比利时医疗数据集"高质量跌倒仿真数据（High Quality Fall Simulation Data）"中的影像素材——该数据集由比利时鲁汶大学（KU Leuven）先进集成传感实验室的研究人员录制，通过模拟跌倒场景来训练跌倒检测算法。依托这类素材，该项目揭示了技术流程与戏剧表演之间令人不安的重合：科学家为机器的凝视演绎假想的虚弱状态。该项目拓展了《机器聆听（Machine Listening）》对数据集如何塑造感知、认知与权力的研究脉络，探讨当护理行为转化为数据点、当身体被建模为衰老、脆弱与死亡的预测模型时，将会产生何种影响。

贡献：该作品由Machine Listening团队（肖恩·多克雷（Sean Dockray）、詹姆斯·帕克（James Parker）、乔尔·斯特恩（Joel Stern））开发，其双频道视频装置将原始研究影像与基于生成式AI（Generative AI）重构的同款场景相结合。音乐家莉齐·韦尔奇（Lizzy Welsh，小提琴演奏）、杰西卡·阿佐迪（Jessica Aszodi，人声演绎）与乔尔·斯特恩（Joel Stern，合成器演奏）为一部本无意面向人类观众的影片创作了现场配乐。该作品探讨了双重失效的命题：人类无法令人信服地演绎虚弱状态，而机器也无法还原这一状态。这双重的不完美成为反思具身性、共情与算法介导式护理的核心场域。

意义：该作品曾展出于《哀悼之后（The Mourning After）》展览（RMIT设计中心画廊，2025年7月24日—9月20日，由拉丽莎·约尔特（Larissa Hjorth）策展），《55次跌倒/环境辅助生活》为在技术重塑的世界中对悲痛、仪式与失落的集体探索贡献了重要力量。该作品将护理重新阐释为人类关系与技术抽象的结合体，揭示了在算法时代，当生命本身成为训练数据时，亲密关系、依赖关系与死亡认知将如何被重构。