iM | Devpost

Heart Disease Detection
COVID-19 Detection
Lung Cancer Detection + Breast Cancer Detection
Chaos Systems and Dynamics Calculator
Outputted Lorenz Attractor w/ Default Input Values (via Chaotic Systems Calculator)
Outputted Lorenz Attractor Data w/ Default Input Values (via Chaotic Systems Calculator)
Outputted Logistic Map w/ Default Input Values (via Chaotic Systems Calculator)

𝐈𝐧𝐬𝐩𝐢𝐫𝐚𝐭𝐢𝐨𝐧

Our overarching idea was sparked by the infamous COVID-19. Not only has COVID-19 killed millions of individuals worldwide, but it has also indirectly harmed our collective well-being as a species. Indeed, vulnerable populations, as well as ill patients, can no longer meet with their doctors at their typical hospital, due to the risks such contact poses in terms of the ongoing pandemic. In addition, physicians can only identify the risk of heart disease with a bewildering and terrifying 20% accuracy, despite being responsible for 32% of all deaths in the world. Not to mention, heart disease is the leading cause of death in America, India, and more. Despite acquiring over 10 million cases per year, and can occur with the subtlest of symptoms that, in reality, require immediate medical treatment. As a result, we noticed a need for a reliable, convenient, and accurate mechanism to correctly self-diagnose patients who potentially suffer from cardiac illnesses, COVID-19 and breast cancer, so that they can take immediate and effective action. We learned about the most common and severe conditions plaguing third-world nations, countless areas that do not have proper access to medical care, and typically find out about major medical issues when its too late. We strove to change this, and allow people to become informed about their health consistently, easily, and effectively. Furthermore, to truly explore the applications and opportunities offered by MecSimCalc, we decided to challenge ourselves with a portion of the project that is rarely done: an all-in-one source for chaotic computation. Presently, there exists little means of accessible computation to chaotic systems. Due to the complex nature of chaos theory, it is often challenging, if not impossible, to find a quick, convenient, and available instrument. Thus, we were provoked to resolve this issue, as for the first time, there finally exists an accessible 3D academic computational tool for chaos theory —specifically— for the Lorenz Attractor and Logistic Map. Chaos theory can be applied to not only surfacing research in medicine, but biology, economics, weather prediction, and much more.

𝐖𝐡𝐚𝐭 𝐢𝐌𝐞𝐝𝐢𝐜 𝐃𝐨𝐞𝐬

With a plethora of practicalities, iMedic enables numerous versatile utilities. From the detection of infectious, cancerous, and cardiovascular diseases using a variety of diverse means; to academic tools such as 3D computations of Chaos Theory, iMedic is a multipurpose platform that uses machine learning, AI, libraries, and big data. As an app created for Third World countries & rural areas, it is for individuals who do not have access to medical support nor fast and efficient examination. iMedic is designed to be used frequently to evaluate the severity of the most common medical conditions, & to provide users with knowledge of their current health. In addition, iMedic includes cutting-edge tools for academic research. By including a chaos dynamics calculator—that computes & graphs chaotic systems such as the Lorenz Attractor & Logistic Map with yielded data via input—iMedic has many applications to fields extending far beyond medicine, such as data science, math, and CS as well.

𝐇𝐨𝐰 𝐖𝐞 𝐁𝐮𝐢𝐥𝐭 𝐈𝐭

Using MecSimCalc, we were able to create a simple UI for users to input their symptoms/data. To complete all the algorithms, we split all the algorithms to complete between our team of 3. One of the team members handled the chaotic system calculator, breast cancer detector, and application interface, another created the ML algorithms for COVID-19 and Heart Disease, and the last member handled the documentation, datasets, and output. For our machine learning algorithms, we used datasets and big data from Kaggle and numerous libraries. Through lots of trial and error, we were to build and test three ML models in a span of one week. For all three ML models, we did lots of preprocessing(especially in the breast cancer detector) to improve accuracy and efficiency. Our code is available to view below.

𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞𝐬 𝐖𝐞 𝐑𝐚𝐧 𝐈𝐧𝐭𝐨

We ran into a few initial challenges throughout this project, especially in terms of the disease detectors. For instance, a notable problem was some models experienced trouble isolating a tumour from the rest of the inputted image's background for the cancer detectors. However, these issues were solved by incorporating increasingly diverse and larger datasets in training our model. Through meticulous trial and error, brute force, and deliberate planning, we managed to work through each issue successfully (with the addition of sacrificing sleep).

𝐀𝐜𝐜𝐨𝐦𝐩𝐥𝐢𝐬𝐡𝐦𝐞𝐧𝐭𝐬 𝐓𝐡𝐚𝐭 𝐖𝐞'𝐫𝐞 𝐏𝐫𝐨𝐮𝐝 𝐎𝐟

Significant accomplishments include the implementation of various datasets into machine learning algorithms that can diagnose cancerous and cardiovascular diseases accurately. Our accuracy rate for our Covid 19 diagnostic algorithm is 97%, and that for our heart attack prediction algorithm is 87%. We also managed to implement a chaos dynamics calculator that computes & graphs chaotic systems such as the Lorenz Attractor & Logistic Map with any inputted data. Finally we managed to implement all of this and its related libraries into MecSim calculator which serves as an effective and aesthetically pleasing front end for our python code.

𝐖𝐡𝐚𝐭 𝐖𝐞 𝐋𝐞𝐚𝐫𝐧𝐞𝐝

We learned to incorporate multiple datasets into a singular machine learning algorithm to increase its accuracy and return the most effective possible diagnosis to our users. We also learned how to incorporate our python machine learning code into mecsim calculator through using the documentation [which was very useful :) ], which enabled our project to have an effective and aesthetic design. Finally, we learned how to conceive an idea, and use our skills and resources to create a viable product.

𝐖𝐡𝐚𝐭'𝐬 𝐍𝐞𝐱𝐭 𝐅𝐨𝐫 𝐢𝐌𝐞𝐝𝐢𝐜

iMedic hopes to grow to be able to revolutionize diagnostic care across areas without medical access. To expand our reach, we plan on integrating mecsim calculator's map feature to allow users to find nearby hospitals and utilize python code to automatically email users occasionally to get a diagnosis and advice on their condition. At iMedic, accuracy is utmost important to us, so we hope to incorporate even more databases into our machine learning algorithms in order to improve our accuracy for all diagnoses, and finally we hope to allow users to import csv files of their past medical history to allow our tools to give them more accurate care.