Name of project
- APPLICATION OF ARTIFICIAL INTELLIGENCE (AI) IN DIAGNOSIS AND RECOMMENDING MEDICINES FROM PATIENT'S HEALTH INDICATORS AND PREcursors of SERIES OF ARTIFICIAL INTELLIGENCE SYSTEMS, DEEP LEARING, BIG DATA DIAGNOSING MANY TYPES OF DISEASES
OJECT SUMMARY
During the last COVID pandemic, our country had more than 11, 600, 000 positive cases and 43, 206 deaths due to COVID. Besides, recently there has been an epidemic of pink eye causing panic in the community.
Every day, caregivers must spend at least 2 times to measure the health parameters of each patient being treated at the hospital. For each patient, caregivers take approximately every 7 minutes. This takes too much time and energy of medical teams.
From there, we hatched the idea of researching to create a product that patients can carry with them to monitor their health and warn their health status in time. In order to minimize the risk of death and ensure the safety of medical staff and doctors in the current epidemic situation.
The solution has two parts: equipment for patients and management for doctors:
The patient section is a device that measures SPO2 indicators, heart rate, temperature, current position GPS and transmits the indicators to the management section for doctors, there are two types of devices: a portable meter and a bracelet.
The management section for doctors has the function of receiving all parameters from patients with devices, statistics measurement results in the form of charts and automatically alerting if parameters exceed the allowed threshold, especially there are two functions of applying artificial intelligence (AI) to automatically suggest drugs and automatically diagnose pneumonia through scans X-rays help doctors who don't need to be on duty still handle them automatically.
The solution has been tested in practice and received positive reviews from doctors at the hospital.
A. REASONS FOR CHOOSING THE TOPIC
The idea came from the fact that a team member had a long period of time caring for a sick family member in the hospital. Although the medical team regularly visits several times a day, family members still die from stroke when the emergency is not timely.
Besides, in Vietnam, 52.8% of people have normal blood pressure (23.2 million people), 47.3% of people (20.8 million people) have hypertension. In particular, among people with hypertension, 39.1% (8.1 million people) were not detected with hypertension, 7.2% (0.9 million people) had untreated hypertension, 69.0% (8.1% million people) had uncontrolled hypertension and every year in the world about 8.5 million people die from high blood pressure, this disease can lead to stroke, heart failure and kidney failure. (1)
During the last COVID pandemic, our country had more than 11, 600, 000 positive cases and 43, 206 deaths due to COVID. Besides, recently there has been an epidemic of pink eye causing panic in the community. (2)
On the other hand, the situation of human resources in 2021 and 2022 in our country only has about 60, 000 medical personnel with additional training, while in fact, this demand needs about 200, 000. Our country is one of the countries that has a serious shortage of human resources in the field of health. (3)
Every day, caregivers must spend at least 2 times to measure the health parameters of each patient being treated at the hospital. For each patient, caregivers take approximately every 7 minutes. This takes too much time and energy of medical teams. (4)
The health of Tien Giang province is still limited in medical examination, diagnosis of lung diseases, new flu variants, and the cost of medical examination and prescription is still quite expensive. (5)
In remote provinces, medical equipment is still rudimentary, doctors in remote areas have not updated with new diseases. (6)
The department's target is AI, digital transformation, this is 1 of the solutions contributing to digital transformation in healthcare. (7)
On average, over 1, 000 people diagnose the disease, at least 25-50 patients are misdiagnosed, leading to death. (8)
From there, we hatched the idea of researching to create a product that patients can carry with them to monitor their health and warn their health status in time. In order to minimize the risk of death and ensure the safety of medical staff and doctors in the current epidemic situation.
B. RESEARCH QUESTIONS, RESEARCH PROBLEMS, SCIENTIFIC HYPOTHESES
Research question: How to reduce manpower for the health sector to avoid being overwhelmed? How can patients check the parameters themselves at home? How can doctors supervise multiple patients at the same time? How can artificial intelligence be applied to warn health indicators, suggest dispensing drugs and diagnose abnormalities on X-rays if doctors are not present in time?
Research problem: Research on equipment and systems for patients to carry on their bodies at all times to monitor parameters
Scientific hypothesis: Help patients feel more secure during treatment at hospitals, minimize the number of unintended deaths as much as possible, increase the productivity of medical staff, support newly graduated doctors in diagnosing diseases. At the same time, create a telemedicine product to be able to examine and treat infectious diseases. To help people have a safer, happier life.
C. DESIGN AND RESEARCH METHODS
1. Subjects of study
The object of our research is systems and devices that patients can wear on their bodies. In addition, our subjects surveyed the demand for using products to support medical examination and treatment for caregivers, nurses, pharmaceuticals,...
We did a market survey (4) of the study subjects, which showed that they spent an average of three hours a day taking patients' health parameters and that more than 97% of those surveyed wanted to experience the group's product.
2. Research contents and tasks
Research and develop medical devices that can be worn by patients to collect health parameters and alert the medical team when patients have worsening cases during treatment. Besides, using artificial intelligence to help the team of medical doctors, especially newly graduated doctors in diagnosing diseases.
3. Research methodology
To carry out the project we used two main research methods:
- Methods of theoretical analysis and synthesis: Theoretical analysis studies different texts and documents on how to produce products, thereby linking each side, each important information part, suitable for the purpose of serving the topic of research and production of our products.
- How to apply theory into practice: Use the foundational knowledge taught at school and the theoretical materials of each part to apply the production of each part of the product and finally assemble them together.
Besides, using statistical methods, to statistics the demand for products in hospitals.
4. Overview of smart health meter
When the patient is admitted to the hospital, it will be provided with a monitoring device, the device plays the role of collecting the patient's health parameters and then sending it to the system, the system Application of artificial intelligence diagnosis and drug suggestion from the patient's health index plays the role of analyzing and processing data, If bad cases occur, the system will automatically alert to the monitor located in the doctor's duty room.
The system called Application of Artificial Intelligence to diagnose and suggest drugs from patient health index brings our desire to convey to patients that always feel secure because there is always a nurse by your side, we will protect you during treatment.
Functions and structure of the system:
a/ Data collection
Use temperature sensors, heart rate, to collect health data from the patient then send it to the system.
b) Preparation tools
The hardware tools that need to be prepared in the test product are:
- ARM Cortex-M4 central processing chip (CPU) core
- Heart rate sensor and blood oxygen MAX30102
- MLX90614 temperature sensor
- Pin: Lithium Battery
- Material: TPU thermoplastic
5. Theoretical basis
a) Arduino
Arduino plays a role in manufacturing the tracking device in the research project. Arduino is an open-source, software-enabled platform that connects to computers to access sensor data outside the world and provide feedback to you. The Arduino will connect itself to each other, connect itself to other devices, connect to the controller chip
b) Python
Python plays a role in building an AI model used to support disease diagnosis. Python is a general-purpose programming language widely used in artificial intelligence development.
c) Winform C#
Winform plays a role in building the application display interface. Winform is a free and graphical library included as part of Microsoft. The.NET Framework provides the foundation for writing other rich applications on the desktop.
d) Private Cloud
Private Cloud is a private cloud computing service often used for businesses to ensure data security. Private Cloud will be protected inside the firewall of the company and directly managed by the enterprise.
6. Design and manufacturing
The solution is designed through two versions: Compact meter and Smart Health Bracelet with functions that can measure heart rate, SPO2, body temperature, patient GPS positioning, thereby sending signals to processing software on computers that use computer vision and artificial intelligence for autoplay medicine.
The solution is designed with drawings on Corel Draw 2018 software, cutting frames and putting components. The kit includes a compact gauge version and a bracelet version for convenience of use.
7. Electrical circuit design
The circuit has an Arduino UNO or Arduino Nano processor, connected to an LCD display, heart rate sensor, SPO2 sensor, current position GPS sensor, temperature sensor.
8. Perfect the solution
The solution after completion consists of two parts: The hardware is a compact meter and bracelet and the software is "Application of artificial intelligence (AI) to automatically dispense drugs based on measured parameters and identify pneumonia through X-rays.
9. Instructions for using the solution
a. AUTOMATIC DRUG DISPENSING FUNCTION
-Step 1: Measure the parameters
Put your hand into the meter or bracelet to measure SPO2 readings, temperature, heart rate, measurement results displayed on the LCD screen, enter those parameters into the software to automatically dispense the medicine, or plug the connection cord between the device into the software to automatically transmit digitally. It is possible to measure the current position, GPS to monitor the patient, in case the patient gets worse, the doctor can provide emergency care in time.
-Step 2: Data processing and storage
Sever plays the role of receiving data sources from the sending device and saving that data source to the cloud database of the system.
-Step 3: Send data to the interface
The system will then access the database system to get the patient's health parameters and draw line graphs according to the data source using Winform.
We select the patient's X-ray image file to be diagnosed. Upload the image file to the display interface.
Then we send the request to sever, the server plays the role of loading the AI model we trained earlier. Obtain the result and return the display interface.
10. About the applicability of the solution
- The solution has technical drawings and has been tested, so it can be mass-produced at low cost, proving AI technology application skills and the completeness of research products.
- The solution has been tested and evaluated well by doctors (see the appendix of the doctor's scientific product evaluation form at the hospital on the last page of the report) can be applied to the health sector inside and outside the province to reduce work pressure for medical staff.
11. About efficiency
- Social Efficiency:
+ Ensure timeliness in medical examination and treatment. Immediate warning when patients have bad cases during their recovery while being cared for in medical facilities.
+ Solve the problem of lack of human resources in the health sector. Caregivers no longer have to spend so much time taking health parameters of each patient. But each patient's health parameters are constantly updated, which caregivers cannot, which reduces errors in the diagnostic process, provides more information and options to optimize treatment pathways, ensures patients receive the best care, and increases resilience. At the same time, ensure a safe distance between health workers and patients from infectious diseases.
+ Help detect signs of illness earlier, thereby warning doctors to treat promptly. This can reduce mortality and improve patients' quality of life.
+ Suggest treatments based on medical data and health indicators of patients. This gives doctors more information and options to optimize the treatment process, ensure patients receive the best care, and increase recovery.
+ By using algorithms and predictive models, AI can make choices and assess the risks of each decision. This helps the doctor get important information to make a final decision and increases the likelihood of an accurate diagnosis.
+ Provide remote diagnostics and drug suggestions, helping to bring health care to remote areas where there are not enough doctors or medical resources. This can improve access to and equity in health care delivery for everyone.
+ Easy to use, easy to access to support children, the elderly, people who are afraid of hospitals, doctors can do it themselves at home, then the family brings the results or videos to the doctor for the doctor to verify, thereby giving advice and appropriate treatments to patients.
- Economic efficiency:
+ The novelty of the solution is to improve the quality of medical examination and treatment so that it can help patients as well as patients' families feel secure in the process of health recovery. Instead of having to search for information and refer to multiple resources, doctors can use AI tools to get immediate access to necessary information and reduce working time. Thereby improving the productivity of the medical team, solving the problem of patient overload in the current situation, the medical team is still lacking a lot.
+ Use AI to support diagnosis and suggest drugs to help reduce work pressure for doctors. Doctors can rely on AI systems to get useful information, supporting the process of making diagnosis and treatment decisions.
+ Optimize the use of medical resources by selecting the most effective and suitable doses, helping to avoid waste and increase treatment performance.
+ Support doctors in diagnosing diseases and suggesting drugs, helping to increase the likelihood of treatment and improve the quality of health care. This can reduce the rate of disease recurrence, help patients recover quickly, and reduce treatment costs.
- Technical efficiency:
+ Ability to process and analyze large amounts of data from patient health indicators to help doctors better understand diseases and create the most accurate analysis for diagnosis and treatment.
+ AI uses machine learning techniques to continuously improve the ability to diagnose and suggest drugs. By learning from data and doctors' feedback, AI can become increasingly smarter and provide better suggestions over time.
12. Novelty, creativity
- Application of both software (artificial intelligence (AI)) and hardware to support medical doctors. AI has the ability to multi-dimensionally analyze the patient's health indicators, from clinical data, medical images, thereby creating a comprehensive view of the patient's health for doctors to diagnose and suggest drugs more accurately.
By using algorithms and predictive models, AI can make choices and assess the risks of each decision. This helps the doctor get important information to make a final decision and increases the likelihood of an accurate diagnosis.
13. Development orientation
- We will continue to research to optimize costs and develop products towards connectivity and remote control, integrating more data so that AI can recognize and suggest drugs for many diseases. In the short term, we will continue to develop towards early warning of cancer through blood tests and identification of malignant or benign brain cancer cells to suggest early treatment for patients.
- Developed so that AI can use data on patients' health indicators to make suggestions on treatment, nutrition, rest, exercise and sports. For example, based on a patient's medical history, drug interactions, and health data, AI can recommend a list of appropriate medications or adjust current drug dosages, and generate tailored personalized treatment regimens for patients based on data from thousands of previous treatment cases. This can provide assistance to doctors in choosing the best treatment for patients, reducing the risk of making mistakes during treatment.
- Enhance diagnostic capabilities through upgrading algorithms and providing larger amounts of data so that AI is capable of detecting potential diseases. AI tools can learn from thousands of medical records and images to recognize disease manifestations and symptoms. This helps provide doctors with supporting information to make accurate and quick diagnostic decisions.
AI can develop medical applications and chatbots to provide online medical consultations to users. Thanks to data analysis and medical knowledge, AI can answer questions about symptoms, provide information about diseases, and make initial suggestions about treatment. This can help users quickly access accurate medical information and reduce the pressure to ask medical providers.
- In addition, we will ask for school-scale testing to get feedback and continue to change.
14. Conclusions
We divided the topic into three phases before bringing the first product to market: Preparation, testing and perfection.
Currently, we are in the first stage, researching and perfecting products, building brands, planning marketing plans, surveying the market to create trust for people when using the product, directly answering the question why people should go to hospitals using the system. It is expected that in the next 1-2 years, we will reach phase 2 to launch the first product model, then collect comments from users. In phase 3, we will handle all user comments and perfect the product in the best way for everyone. With the cost of the tracking device, we estimate about 1 million VND and sell it to the market for about 1.5 to 2 million VND. With an extremely competitive price in the health equipment market and applied to private hospitals.
In the journey to accompany the project, we realize that wherever people are, they care about their health and their loved ones, please accompany us to reduce the number of deaths in the future as much as possible. For everyone to have a safer life. We are committed to accompanying the project so that we can help the most people in need. Thank you very much.
E. REFERENCES
1. Arduino.vn
2. Heart rate sensor and spo2, arduino.vn
3. Sensor measuring temperature, Arduino.vn
4. GPS current location, Arduino.vn
5. Machinelearningcoban.com
Comments