Machine Learning in Healthcare for Better Diagnosis & Prognosis
Sri Deepthi Chava, Vemparala Soumya Bhuvana Priya, Vemulamada Chandrasekhar, Chinthala John Moses, Vijaya Babu Burra, Prasanth Yalla
dept. of Computer Science & Engineering
Koneru Lakshmaiah Education Foundation
Vaddeswaram, India
Abstract— A diagnosis is the determination by a medical expert of the illness, problem, or condition affecting a patient. Research on the patient, such as physical examinations or medical tests, are typically necessary for a diagnosis. The phrase can be used to describe both the decision-making process and its outcome. Identifying the precise source of your ailment is the end goal in either case. In medicine, a prognosis is a forecast of the anticipated effects of a condition on a patient. The phrase refers to a forecast of how likely or unlikely a patient's recovery is, and is often used in the context of more severe diseases and ailments (such as cancer). Modern hydraulic unit health diagnostic systems are crucial for maintaining the hydroelectric power plant's dependability and safety (HPP). However, they are unable to give prompt identification of operational flaws as severe as fatigue cracks. The two primary causes of this issue are discussed in this article. The first is that hydraulic units have a high degree of uniqueness, which prevents the successful application of statistical information processing techniques, such as big data and machine learning technologies. The second is that it is fundamentally impossible to detect fractures in several important hydraulic unit components alone by using data analysis from a typical diagnostic system at the HPP. This was supported by developed computational analyses using the Francis turbine as an example. A prognostic block for an individual analytical prognosis of the unit's residual lifespan based on the computed evaluation of fatigue strength is suggested as an addition to the capabilities of conventional diagnostic systems. The conceptual diagram and demonstration version of the suggested analytical predictive system are presented in this article. The advantages of the suggested method are demonstrated by an evaluation of the usual vibration diagnostic both the suggested technique as a device for the quick identification of fractures in a Francis turbine runner.
Keywords— machine learning; artificial neural networks; decision tree; support vector machine; k-nearest neighbor