Biologically informed neural network for subphenotype classification in septic AKI
(2022) BMEM01 20222Department of Biomedical Engineering
- Abstract (Swedish)
- Sepsis is a life threatening condition where the body’s reaction to an infection results in a dysregulated immune response - ultimately causing damage to tissues and organs. The syndrome is diverse, both in underlying biology, disease manifestation and severity, and is therefore divided into endotypes and further into subphenotypes. Further understanding of the biological pathways of the various sepsis types is required in order to develop targeted diagnostic and therapeutic tools necessary to combat the disease. In this thesis, the plasma proteome of patients suffering from two subphenotypes of septic acute kidney injury with varying severity were analyzed. The proteomic data was combined with the Reactome pathway database, and leveraged... (More)
- Sepsis is a life threatening condition where the body’s reaction to an infection results in a dysregulated immune response - ultimately causing damage to tissues and organs. The syndrome is diverse, both in underlying biology, disease manifestation and severity, and is therefore divided into endotypes and further into subphenotypes. Further understanding of the biological pathways of the various sepsis types is required in order to develop targeted diagnostic and therapeutic tools necessary to combat the disease. In this thesis, the plasma proteome of patients suffering from two subphenotypes of septic acute kidney injury with varying severity were analyzed. The proteomic data was combined with the Reactome pathway database, and leveraged to generate and train a biologically informed neural network in classifying the two subphenotypes. The network was able to distinguish between the subphenotypes, achieving an accuracy of $98.2 \pm 0.02\%$ when created with four hidden layers. The informed nature of the network allows for introspection into the network's decision making - allowing us to utilize feature importance values to interpret which proteins and biological pathways the network deemed important for classification. Ultimately, this identified several biomarkers for the subphenotypes including apolipoproteins, histones and known inflammatory markers such as CD14 and osteopontin. The algorithm generating the biologically informed network was generalized and is publicly available as a Python package: https://github.com/InfectionMedicineProteomics/BINN. (Less)
- Popular Abstract
- Utilizing machine learning to understand sepsis
Sepsis is one of the deadliest syndromes in modern time, with little to no effective therapies available. In this project, machine learning was utilized to gain insight into the underlying biology of sepsis - a necessary step in finding novel and effective treatments and diagnostic tools.
Sepsis is a syndrome (collection of symptoms) which is responsible for about 20% of global deaths each year. It is extremely diverse and complex, rendering it difficult to both diagnose and treat. Recently, researchers have classified various types of sepsis and recognized that unique therapies are required for the different types. However, before creating treatments and diagnostic tools, we need to... (More) - Utilizing machine learning to understand sepsis
Sepsis is one of the deadliest syndromes in modern time, with little to no effective therapies available. In this project, machine learning was utilized to gain insight into the underlying biology of sepsis - a necessary step in finding novel and effective treatments and diagnostic tools.
Sepsis is a syndrome (collection of symptoms) which is responsible for about 20% of global deaths each year. It is extremely diverse and complex, rendering it difficult to both diagnose and treat. Recently, researchers have classified various types of sepsis and recognized that unique therapies are required for the different types. However, before creating treatments and diagnostic tools, we need to understand the underlying biology of the various types - which is easier said than done.
In this project, machine learning was utilized to understand the biology of a specific type of sepsis which is characterized by damage to the kidney (referred to as septic acute kidney injury or AKI). Machine learning is a way to make a machine find patterns in complex data - and sepsis as a disease can be seen as a complex mixture of biological molecules. Finding the pattern in this soup of molecules is therefore a task fit for machine learning, and could help us understand the disease.
An algorithm named a biologically informed neural network - that is: a machine learning algorithm (specifically a neural network) that reflects the underlying biology of the disease (hence biologically informed). Creating such an algorithm solves the black box problem in machine learning, which states that it is impossible to understand what a machine learning algorithm is doing when it is solving a problem. It also allows for introspection into the algorithm and understand what parts of the biology it finds important and interesting when analyzing specific types of sepsis.
The algorithm allowed for the finding of proteins and biological pathways which were important in classification of septic acute injury of different severity. It was generalized to be compatible with any type of condition, disease or syndrome. Further, the algorithm is available in a public repository, and anyone can now create a biologically informed neural network with one line of code. Therefore, it can be utilized in other experiments to not only progress in the development of treatments and diagnostics of sepsis - but also other diseases. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9103197
- author
- Hartman, Erik LU
- supervisor
- organization
- alternative title
- Biologiskt informerat neuralt nätverk för klassificering av subfenotyp i septisk AKI
- course
- BMEM01 20222
- year
- 2022
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Machine learning, neural networks, proteomics, sepsis, AKI
- language
- English
- additional info
- 2022-21
- id
- 9103197
- date added to LUP
- 2022-11-21 13:07:44
- date last changed
- 2022-11-21 13:07:44
@misc{9103197,
abstract = {{Sepsis is a life threatening condition where the body’s reaction to an infection results in a dysregulated immune response - ultimately causing damage to tissues and organs. The syndrome is diverse, both in underlying biology, disease manifestation and severity, and is therefore divided into endotypes and further into subphenotypes. Further understanding of the biological pathways of the various sepsis types is required in order to develop targeted diagnostic and therapeutic tools necessary to combat the disease. In this thesis, the plasma proteome of patients suffering from two subphenotypes of septic acute kidney injury with varying severity were analyzed. The proteomic data was combined with the Reactome pathway database, and leveraged to generate and train a biologically informed neural network in classifying the two subphenotypes. The network was able to distinguish between the subphenotypes, achieving an accuracy of $98.2 \pm 0.02\%$ when created with four hidden layers. The informed nature of the network allows for introspection into the network's decision making - allowing us to utilize feature importance values to interpret which proteins and biological pathways the network deemed important for classification. Ultimately, this identified several biomarkers for the subphenotypes including apolipoproteins, histones and known inflammatory markers such as CD14 and osteopontin. The algorithm generating the biologically informed network was generalized and is publicly available as a Python package: https://github.com/InfectionMedicineProteomics/BINN.}},
author = {{Hartman, Erik}},
language = {{eng}},
note = {{Student Paper}},
title = {{Biologically informed neural network for subphenotype classification in septic AKI}},
year = {{2022}},
}