Created by: Fan Gao, Chang Liu , Coral Cohen
PBL Design: Chronic disease risk reduction
Introduction to your assignment
Guideline for students
A great example
1. Introduction to your assignment
Chronic illnesses are a major risk for people's lives and they are very common, especially to elder people. According to a US Economist report, chronic illness accounts for more than 86% of deaths in EU countries. In Israel, approximately 21% of the population suffers from a chronic illness and approximately 65% of deaths in Israel are caused as a result of a chronic illness.
In addition, it can be observed that 40% of the population over the age of 15 suffer from chronic diseases and the percentage increases with age. The Economist's report suggests that communities and health systems need to allocate more resources to chronic disease treatment programs in order to reject or prevent the stages of human life that are at risk.
For example, according to Canary Foundation, cancer strikes at the US, one in three woman and one in two men in the US. More than 560,000 people die from it each year. The best chance to reduce these numbers it through early detection and intervention.
You need to plan PBL to help people identify chronic diseases in early stage and help with an appropriate treatment. The project must be achievable, innovative, simple to operate and based on scientific and engineering knowledge.
2. PBL Requirements
Analyze the existing problem and solutions with adequate academic references and critical thinking via different aspects like professional, physiological, marketing, etc.
Working in small groups to come up with innovative solutions that are practical and expected by the market.
In small groups, every student is supposed to contribute to the project through their own way. The team should be well structured and sub-assignment should be distributed reasonably.
The solution must incorporate scientific or engineering principles from different STEM subjects to develop interdisciplinary abilities.
The final outcome of the solution should be a functional product, model or prototype, with a document about marketing analysis and cost management, etc. All the demonstration needs to be uploaded to Augmented-world.
Offer high-level peer assessment and self-assessment about the work from self, group members and members from other groups, via innovation, workload, feasibility, etc.
3.1. Guidelines for Teachers
Divide students into small groups (3-5) from different majors.
Provide an introduction to the real-world problem and related references.
Give several examples of existing solutions and products.
Introduce basic STEM principles and study method which might be used in the program.
3.2. Guidelines for Students
Group icebreaking, leadership decision and group working structure;
Submit a report that analysis existing problem and solutions with adequate academic references and critical thinking via different aspects like professional, physiological, marketing, etc. [Assignment 1]
The group come up with an innovative solution for the existing problem and generate a report to clearly illustrate the whole idea and program schedule plan.[Assignment 2]
Teacher and TAs give feedback about [Assignment 1 and 2] and give instructions about the design and building part of the product.
Working on the program based on the schedule plan, at the midterm, students need to give a presentation about the progress they have made, a semi-finished model or a product is required.[Assignment 3]
At midterm presentation, students need to assess themselves works, their classmates’ works. Also, the teacher should assess every student’s work from the team and individual perspective.
Optimize the solution according to the assessments and continue finishing the program according to schedule, marketing issues should be involved. Write a report about the whole product with video or image illustration, upload all the material to Augmented-World.[Assignment 4]
Final bidding. All groups should present their final product to all the investors (other students and teachers). And receive assessments from all the investors.[Assignment 5]
4. Assessment System
4.1. Assessment Principles (how to evaluate PBL work)
The topic must balance both “problem solving” and “market situation”, which reflects professionalism. The topic focuses on the important problems, phenomena and development trends in the practice of real life. It reflects the team's ability to comprehensively apply the theoretical knowledge, methods, and techniques of innovation ability to solve practical problems.
4.1.2 Research content
On the basis of grasping the research status and development trend at home and abroad, this project conducts theoretical analysis and research on research issues and draws its own product(prototype) findings and conclusions through reasonable framework and structure.
4.1.3 Innovate methods
The innovate methods should match the life questions, be reasonable, standard and scientific, and encourage the attempt and absorption of new methods and technologies.
4.1.4 Product results
The product results have certain innovation and research findings, research results for the practice of news dissemination have the application or reference value. The document should be upload on the augmented-world website.
4.2 Assessment Criteria
Assignment 1 [10% by the teacher]
Assignment 2 [10% by the teacher]
Assignment 3 [20%, 5% from self, 5% from peers, 10% from teacher]
Assignment 4 [20% by the teacher]
Assignment 5 [30%, by all investors]
Give high-quality assessments to self and others[10% by the teacher]
The evaluation conclusion is divided into four kinds: excellent, good, pass and fail. Excellent: >=90; Good: 89-75; Pass: 74-60; Fail :<60.
5. A Great Example: Sticky Guard
In light of the horrendous deaths caused by chronic illnesses and with inspiration from the recommendations for early detection that we showed at the first paragraph, in order to receive proper treatment before it is too late, especially for the old, we have devised an innovative invention that will make it easier for the public to monitor their health condition, alert them to potential risks and how to treat them before they develop into severe diseases.
Our invention consists of small, flat and sticky sensors that can be pasted under a wristwatch or bracelet, and even above the skin, concealed with a make-up. The sensor is linked with a Bluetooth system to transfer heart pulse date to a dedicated application. In the application, the data information would be processed in an artificial intelligence system via deep learning technology, which could compare users' data with well-trained disease models including diabetes, high cholesterol, high blood pressure and sleep apnea. The application would alert you the chances of having a chronic illness, what percentage the risks is and what the suggested actions are.
Why choosing our product is the best for you?
Today, many of the chronic illnesses can be discovered through extensive hospital examinations, but hospital admissions are not sufficient to get such a large number of patients in a short time. In addition, a large number of tests require the public to lose several working days and wait long queues, which means that they do not schedule medical tests in the first place and the cause in known - they detect serious illnesses only when the situation is too late and they are in real life danger. Therefore, our invention does not require the public to go through many tests but to examine their health condition effortlessly - all it takes is a dedicated sticker and a mobile phone of the subject or our family members who receive the information related to the risks relevant to the person being exposed.
Furthermore, a market survey showed that our invention is unique - no similar inventions exist.
We think our invention is very innovative because it can save lives without making a person obsessed with the hospital testing, without carrying an invasive or inconvenient device. Our sticker can be pasted on bracelets, watches and flying on bare skin without feeling it – a revolutionary solution. Older people who have difficulty understanding apps can also cope with its simple operation and even have the ability to automatically transfer information to family members' smartphones and for that reason – the family can follow the health condition of their beloved ones any time.
6. Scientific Background
Our main goal is to detect chronic diseases in advance through wearable devices. To realize our goal, we need some background knowledge in biology, electrical engineering and computer science. We will briefly describe the process about how to implement this idea. The first step is to use a photosensitive sensor to detect your heart rate. After we achieve the data, we will transfer it to the phone via bluetooth module. The application designed by us will be running on the mobile phone. The app keeps receiving new data from bluetooth and input it into the neural network, and return with results about the potential risk level to tell whether the user has a specific chronic disease or not.
6.1 Sensor Part
The device will contain four main modules: a green LED light, a photosensitive sensor, a bluetooth module and a power supply. We will use PPG (Photo Plethysmo Graphy) method to monitor the heartbeat. The LED light and the photosensitive sensor will be used to monitor periodic changes in the light intensity of blood flowing through the wrist. Green LED lights are used because the absorption of green light is the greatest in the presence of the red blood. This will make our detection result more accurate. The volume of blood in the skin pulsates with beats of the heart. When the heart contracts, the green light is absorbed maximumly and the light intensity is the least. When the heart diastole, green light is absorbed minimumly and light intensity is the greatest. This allows continuous monitoring to detect periodic changes in the intensity of light, which can then be converted into electrical signals to calculate the user's heart rate. After getting the data, we will transfer the pulse data to the app on the phone through the bluetooth module, which brings us to the second part.
6.2 Application Part
Designing applications on mobile phones mainly involves two parts: app development and algorithm design. As for the application development, we hope this app can be cross-platform in order to have more potential users. So we will use the Flutter framework developed by Google and write our program using Dart and Java. This will make our application executable on both iOS and Android platform.
Algorithm design is the core of our device. Our goal is to calculate whether a user is at risk of developing a chronic disease through continuous heart rate input. In the past few years, neural networks have proved to be a powerful tool. On the mathematical level, with enough neurons and hidden layers, neural networks can fit into any function with high accuracy. So we're going to build a RNN (Recurrent Neural Network). This kind of neural networks are good at processing sequential data.
The only input of the neural network is the time series about the heartbeat. The output is the multi-classification result processed by softmax function. It will give the probability of various chronic diseases. Considering that we need to analyze a time series with great length, this will usually lead to the problem of gradient explosion/vanishing of the neural network. To avoid this situation, we will adopt special gate structures such as LSTM and GRU to prevent the forgetting of the past data. We will use GRU to save limited computing resources on mobile phones. Compared to the LSTM technique, GRU has lower performance requirements and can maintain nearly the same accuracy in the meantime.
After designing the structure of neural network, we need dataset to train the neural network. We need about 10 devices to collect data or to use public datasets directly. After training the neural network, we save the parameters of the network structure on the phone. The time series data of the user's heartbeat will be input into the process of forward computation of the neural network to get the probability of whether the user has any chronic diseases. When the probability exceeds the thresholds we set, the app will display users with different risk level tags and corresponding suggestions.