Request a Meeting
Media Coverage

Malaysia Sin Chew Daily: SmartPeep uses AI to predict and prevent falls for the senior safety 星洲日报:重视老人行动安全 SmartPeep用AI防跌

Our CEO & Founder Lim Meng Hui (Ph.D) was interviewed to explain how our fall prevention and fall detection systems work to support the safety of the elderly. Highlighting his mission in aiming better quality care and efficiency in nurses workflow.

Posted on
September 2019
More articles

Singapore-based startup SmartPeep, has developed an AI Patient-Sitter System in 2016 and installed it in hospitals and nursing homes to immediately alert nurses before patients fall out of bed to prevent repercussions.

The system uses AI algorithms to detect and analyse patient movements 24-hours a day through optical algorithms. Once the other patient moves or gets out of bed, the nurse will be notified in two ways: first, through the SmartPeep control panel; second, through the mobile phone application. "Our goal is to call the nurse before the patient gets out of bed."

Currently, the AI Patient-Sitter System is a starting point. SmartPeep would like to develop a multipurpose system. For example, the AI system will be able to track the elderly activities and analyse their behaviour. Furthermore, for the bed bound patients, it may also develop a "reminder" feature to alert nurses or caregivers to turn the patients on time. 

Lim Meng Hui (Ph.D), who is good at face recognition technology, also pointed out that it is possible to integrate face recognition technology into the AI System in the future, but SmartPeep will be more focusing on preventing falls for the time being.

However, SmartPeep AI System is not open to consumers yet. The main focus at this stage is on hospitals and nursing homes.

Refer to the original article: Click here.

年纪越长越怕跌倒,不时听闻高龄者不慎跌倒,轻则扭伤或擦伤;重则骨折、头颅出血或甚至夺命。跌倒已是银发族日常生活中一大隐患,有些也因跌倒而心生恐惧,不愿走动太多。我国初创公司SmartPeep在2016年研发出人工智能(AI)看护系统,并安装在医院和疗养院,在病患者跌下床之前能迅速向医疗人员发出警报,避免任何悲剧发生。

“从2018年到现在,我们的系统已经预防超过500宗跌倒事件。”SmartPeep联合创办人林铭辉自豪的说。他过去曾向不少医院和疗养院单位了解,尝试找出老人或病患常跌倒的原因,当中有部分例子是患有失智症老人,由于行动不便,很常因下床跌倒;有些则不会使用医院紧急求助系统(call bell system)请求护士帮忙,而在床边或厕所里摔倒。

2017年,他们终于有了产品的雏形,并在北马一家专科医院测试,有关测试也花了6个月时间。不过,测试结果却令他们发现系统有很多缺陷需要补足。

说回那一次的测试,林铭辉大叹一声说,太多意料之外的事发生。他们自身试用,准确度可达到94%。“当我们第一天在医院试用时,8张病床突然有千多个虚假警报。”这并非系统之过,源自于他们当初设计系统时,只为了应对一个病患的动作,却没设想到病患的家属会来探访,多人一起互动时就会让系统“失误”。得知原因后,他们再用两三个月时间重新训练机器,最后达到95%的准确度。

有关系统会借由光学传感器通过AI运算法,24小时不间断的检测和分析病患动作。一旦对方有移动或下床的动作,会以两种方式通知护士,第一、通过SmartPeep的求助系统;第二、通过手机应用程式。“我们的目标是在病人下床以前,就要先把护士叫过来。”

目前,这套AI智能看护系统只是一个起点,SmartPeep想开发更多元用途的系统,比方说未来两年会通过闭路电视检测病患或老人的日常活动(activity recording),从中分析他们的行为举止是否有异或突然改变。又或者,针对长期卧在病床的病人,他们每两小时得翻身一次,避免背部皮肤溃烂。看护系统或许会设计一个“提醒功能”,让护士或看护人员准时帮助他们翻身。擅长人脸识别技术的林铭辉更指出,未来也有可能在看护系统置入人脸识别技术,不过现阶段会着重在预防跌倒。

不过,林铭辉称,如今有关系统仍未公开给消费者使用,“如果是住家,还得考量光线不足、会不会有家私遮挡等因素,现在推出市场恐怕会遇到很多问题。”这两年他们则会专注在医院和疗养院。

参考原始文章: 点击这里

More articles