Ensemble Learning Improvement of Clinical Diagnoses of Knee Osteoarthritis Risk in Adults
Olayemi Olufunke Catherine,
Olasehinde Olayemi Oladimeji,
Alowolodu Olufunso Dayo,
Osho Patrick Olarewaju
Issue:
Volume 11, Issue 4, August 2022
Pages:
51-64
Received:
28 May 2022
Accepted:
27 June 2022
Published:
26 July 2022
DOI:
10.11648/j.ijiis.20221104.11
Downloads:
Views:
Abstract: Background: Knee Osteoarthritis (KOA) is a deteriorating disease that affects human knee joints leading to impaired quality of life with no curative treatments. Timely detection of KOA will guarantee its good management, prevent cartilage impairment and reduce its rate of progression. To heighten its early detection. Objective: This study developed a machine learning ensemble model that improves early clinical diagnosis of the risk of KOA in Adults. Method: The diagnostic results of three machine learning diagnostic models were combined with two ensemble methods proposed to improve the diagnosis of KOA risks. KOA patient dataset used for the modeling of the diagnostic models was obtained from the Federal Medical Hospital located in Ido-Ekiti, Nigeria. Results and Conclusion: The diagnostic result of the base diagnoses models shows higher accuracy than similar recently reviewed research in the literature. Diagnoses results of the two ensemble models confirm their abilities to improve the results of the base models. From the comparison of the diagnoses of the ensemble methods, the Multi Response Linear Regression model leads with 97.77% followed by the Majority Voting model with 96.54% diagnostic accuracy. The Statistical tests employed in this study, validated the ranking of the results recorded by each of the diagnostic models.
Abstract: Background: Knee Osteoarthritis (KOA) is a deteriorating disease that affects human knee joints leading to impaired quality of life with no curative treatments. Timely detection of KOA will guarantee its good management, prevent cartilage impairment and reduce its rate of progression. To heighten its early detection. Objective: This study developed...
Show More
Robust Lane Detection by Removing Overlapped Objects for Complex Driving Scene Analysis
Rasheed Raed,
Abu Hadrous Iyad
Issue:
Volume 11, Issue 4, August 2022
Pages:
65-69
Received:
18 August 2022
Accepted:
13 September 2022
Published:
11 October 2022
DOI:
10.11648/j.ijiis.20221104.12
Downloads:
Views:
Abstract: Human mistake is virtually always to blame in situations involving motor vehicles, which can have fatal consequences. When it comes to lane detection, the analysis of driving scenarios that is carried out by dashboard cameras that are placed in vehicles as part of advanced driver assistance systems (ADAS) is of the utmost significance. The initial developments in lane detection systems resulted in the creation of two distinct varieties. Image processing and deep segmentation have typically relied on a number of different methods. The techniques of deep segmentation are not yet capable of resolving many of the most important and challenging issues. We came up with a solution to the problem of object lanes overlapping each other and developed a dependable technique for lane detection that can be used in driving scene analysis systems. The method that is provided for real-time object detection makes use of the real-time object detection algorithms that are the most up-to-date and effective currently available; these algorithms are collectively referred to as YOLOv5. By identifying the object-lane that is overlapping the lane that has been unequivocally found by removing items that have overlapped, it is possible to solve this problem.
Abstract: Human mistake is virtually always to blame in situations involving motor vehicles, which can have fatal consequences. When it comes to lane detection, the analysis of driving scenarios that is carried out by dashboard cameras that are placed in vehicles as part of advanced driver assistance systems (ADAS) is of the utmost significance. The initial ...
Show More
