EFFECTIVE FEATURE ENGINEERING TECHNIQUES FOR HEART DISEASE PREDICTION USING ML

Abstract

There have been tremendous developments in
medical services (MS) throughout the years,
with an emphasis on bettering response times,
patient care, and overall results. A
revolutionary strategy to improve emergency
medical treatment is the incorporation of
human interaction technologies and Machine
Learning (ML) into ambulances. Historically,
medical staff have had to rely on rudimentary
tools and clinical judgement to assess patients'
status during crises, and then transmit their
findings to hospitals verbally or in writing.
Although this method has been successful in
the past, it has several drawbacks now,
including transmission delays, human mistake
in reading vital signs, and an absence of
established evaluation criteria. We propose an
integrated system that uses explainable humancomputer interaction to connect hospital-based
decision-making with on-site ambulance
treatment in order to improve emergency
healthcare. Two applications, one for the
hospital server and one for the ambulance,
were developed using Python and Tkinter,
respectively, to form the proposed system.
Through a secure socket connection, the
ambulance app transfers patient data from
stored datasets to the hospital server. On the
back end, the data is prepared for machine
learning algorithms like RFC and K-Nearest
Neighbors (KNN) to provide real-time
predictions about patients' illnesses. The
results are solid and easy to understand since
RFC outperforms KNN with near-perfect
accuracy, while KNN is simple and easy to
understand. Transforming traditional manual
systems into an efficient, dependable, and
transparent AI-driven emergency response
mechanism, this intelligent framework
automates data processing and prediction and
provides visual feedback through explainable
metrics and confusion matrices. As a result,
healthcare is being redefined