Background and Objectives

Type-2 diabetes is one of the chronic diseases. This disease can be controlled by adjusting the dose of medicine, which is calculated from regular monitoring of blood sugar level. Blood glucose estimation methods are grouped into two categories direct and indirect. The direct method (invasive in nature) provides more accurate results; but people are not interested to test their blood several times in the day; because blood sample collection process is painful. On the other hand, indirect estimation methods are popular due to its non-invasive nature. The most widely used non-invasive blood glucose estimation method is based on urine sugar level estimation. Urine sugar level estimation is a chemical process requiring manual involvement. Human nature is very different; they dislike the repetitive work of testing urine regularly, although the process is not at all cumbersome. It will be very helpful if a system exists, which monitors urine sugar level automatically from the toilet.

Methods

This work describes an automatic technique to estimate blood sugar level from urine. The contribution of this work is as follows:

•

A complete customized mechanical unit, which controls the chemical process of urine sugar estimation.

•

An automatic technique to build the fuzzy membership functions from training data set.

This system includes a chemical process control along with a fuzzy logic based color estimation technique, where fuzzy membership functions are derived from training data set. One salient feature of this fuzzy membership functions generator is that it is tuneable, that means it allows calibration after constructing membership functions. From application point of view, it is an intelligent toilet to keep track of blood sugar level from urine.

The system is divided into two sub sections named as a control section and a computation section. The control section includes the control of mechanical units and chemical process initiation. The activeness of chemical reagent changes over time, this system has the provision to handle such situation through volume adjustment chamber. The control section includes a lot of valve control, they are interdependent. Petri-net is used to synchronise them. Computation section is used for estimation of urine sugar level from the changed color of Benedict's Qualitative Solution.

Result

From operational point of view, this system is a combination of sequential and parallel sub processes. It can be divided into 9 sub processes. The time required to complete all 9 processes is 660.5 second. This time includes sample collection time, chemical reaction time, result calculation and system cleaning time. The average Sensitivity, Specificity and error rate of the system are as follows 88.0225%, 95.95% and 5.765%. PIPEv4.3.0 is used to analysis the Petri-net. As per the analysis report, the system is safe (reliable).

Discussion

This system is efficient to estimate blood sugar level from urine. This system senses the urine sugar level indirectly using the color sensor. The color sensor is not directly in touch with the chemical of the reaction chamber. The normal toilet cleaning (acidic) solution can be used to clean the chambers. So, maintenance process is quite easy. The proposed system can reduce the probability of glaucoma, kidney problem etc. by assisting doctors to control high blood sugar level through regular monitoring of urine sugar level.

中文翻译：

---

智能马桶系统，用于无创估算尿液中的血糖水平

背景和目标

2型糖尿病是一种慢性疾病。这种疾病可以通过调整药物剂量来控制，该剂量是通过定期监测血糖水平来计算的。血糖估计方法分为直接和间接两类。直接方法（本质上是侵入性的）可提供更准确的结果；但是人们对一天中的几次血液测试都不感兴趣。因为采血过程很痛苦。另一方面，间接估计方法由于其非侵入性而很受欢迎。最广泛使用的无创血糖评估方法是基于尿糖水平评估。尿糖水平估算是一个化学过程，需要人工干预。人性是非常不同的。他们不喜欢定期测试尿液的重复工作，尽管这个过程一点也不麻烦。如果存在一个系统，该系统可以从卫生间自动监测尿糖水平，这将非常有用。

方法

这项工作描述了一种自动技术，可以从尿液中估计血糖水平。这项工作的贡献如下：

•

完整的定制机械单元，可控制尿糖估算的化学过程。

•

一种从训练数据集构建模糊隶属函数的自动技术。

该系统包括化学过程控制以及基于模糊逻辑的颜色估计技术，其中模糊隶属度函数是从训练数据集中得出的。这种模糊隶属函数生成器的一个显着特征是它是可调整的，这意味着它可以在构造隶属函数后进行校准。从应用的角度来看，它是一种智能尿液，可跟踪尿液中的血糖水平。

该系统分为两个子部分，分别称为控制部分和计算部分。控制部分包括机械单元的控制和化学过程的启动。化学试剂的活性随时间而变化，该系统具有通过容积调节室处理这种情况的功能。控制部分包括许多阀门控制，它们是相互依存的。Petri-net用于同步它们。计算部分用于根据本尼迪克特定性溶液颜色的变化来估计尿糖水平。

结果

从操作的角度来看，该系统是顺序和并行子过程的组合。它可以分为9个子过程。完成所有9个过程所需的时间为660.5秒。该时间包括样品收集时间，化学反应时间，结果计算和系统清洁时间。系统的平均灵敏度，特异性和错误率分别为88.0225％，95.95％和5.765％。PIPEv4.3.0用于分析Petri网。根据分析报告，该系统是安全的（可靠的）。

讨论区

该系统可有效估算尿液中的血糖水平。该系统使用颜色传感器间接感应尿糖水平。颜色传感器不直接与反应室的化学物质接触。普通的马桶清洁（酸性）溶液可用于清洁马桶。因此，维护过程非常容易。所建议的系统可通过协助医生通过定期监测尿糖水平来控制高血糖水平，从而降低青光眼，肾脏问题等的可能性。