Web based Ecosystem Software using PAC Algorithm for Virtual Crossmatching in Transplants

Author(s):

Dona Deric , MARIAN ENGINEERING COLLEGE; Jose Hormese, MARIAN ENGINEERING COLLEGE; E. Arun, MARIAN ENGINEERING COLLEGE

Keywords:

Virtual Cross Match, HLA Epitopes, Web based Software, PAC Algorithm, Computational Learning

Abstract:

Patients who face immunological risks during transplants face difficulties in obtaining the matching donor. Evaluation of this risk is done based on donor-recipient compatibilities. Previous papers made use of EpVix software that performs virtual cross matching and reactivity analyses of epitopes. The best donor can be identified using this software. Patients with high immunological risk have a smaller chance of receiving a transplant which in turn reduces the number of available compatible donors. This paper introduces a computational learning algorithm called a PAC learning which provides more error tolerance facilities and increases the capability of selection among various available donors of the group. The probability of success in selecting the donor increases as PAC learning is noise tolerant in nature. PAC refers to Probably Approximately Correct learning where samples are received and a hypothesis is to be formed from a group of possible donors available. The main advantage is that with high probability, the selection process will have low error. The current issue of the available software is how the selection process is to be conducted to correctly identify the best matching donor when multiple donors are available for particular recipient. Error tolerance has to be increased along with proper filtering to eliminate the ones with lower matching probabilities and finally determine the perfect match for transplant. PAC algorithm solves this issue to an extent.

Other Details:

Manuscript Id	:	IJSTEV3I9143
Published in	:	Volume : 3, Issue : 9
Publication Date	:	01/04/2017
Page(s)	:	294-297

Download Article