This thesis presents the development and evaluation of a computerized physician-parent decision support tool for a neonatal intensive care unit (NICU), known as Physician and Parent Decision Support (PPADS). The NICU is a specialized hospital unit that treats very-ill neonates. Many difficult care decisions are made daily for this vulnerable population. The PPADS tool aims to augment current NICU decision-making by helping parents and physicians make more informed decisions, improving physician-parent communication, increasing parent decision-making satisfaction, decreasing conflict, and increasing decision efficiency. The development of the PPADS tool followed a five-step methodology: assessing the clinical environment, establishing the design criteria, developing the system design, implementing the system, and performing usability testing. Usability testing of the PPADS tool was performed on neonatologists and on parents of neonates who have graduated (survived) from a tertiary level NICU. The usability testing demonstrated the usefulness and ease of use of the tool.

Semantic web based decision support systems represent domain knowledge using ontologies that capture the domain concepts, their relationships and instances. Typically, decision support systems use a single knowledge model—i.e. a single ontology—which at times restricts the knowledge coverage to only select aspects of the domain knowledge. The integration of multiple knowledge models—i.e. multiple ontologies—provides a holistic knowledge model that encompasses multiple perspectives, orientations and instances. The challenge is the execution-time merging of multiple ontologies whilst maintaining knowledge consistency and procedural validity. Knowledge morphing aims at the intelligent merging of multiple computerized knowledge artifacts—represented as distinct ontological models—in order to create a holistic and networked knowledge model. In our research, we have investigated and developed a knowledge morphing framework—termed as OntoMorph—that supports ontology merging through: (1) Ontology Reconciliation whereby we harmonize multiple ontologies in terms of their vocabularies, knowledge coverage, and description granularities; (2) Ontology Merging where multiple reconciled ontologies are merged into a single merged ontology. To achieve ontology merging, we have formalized a set of semantically-defined merging criteria that determine ontology merge points, and describe the associated process-specific and knowledge consistency constraints that need to be satisfied to ensure consistent ontology merging; and (3) Ontology Execution whereby we have developed logic-based execution engines for both execution-time ontology merging and the execution of the merged ontology to infer knowledge-based recommendations. We have utilized OWL reasoning services, for efficient and decidable reasoning, to execute an OWL ontology. We have applied the OntoMorph framework for clinical decision support, more specifically to achieve the dynamic merging of multiple clinical practice guidelines in order to handle comorbid situations where a patient may have multiple diseases and hence multiple clinical guidelines are to be simultaneously operationalized. We have demonstrated the execution time merging of ontologically-modelled clinical guidelines, such that the decision support recommendations are derived from multiple, yet merged, clinical guidelines such that the inferred recommendations are clinically consistent. The thesis contributes new methods for ontology reconciliation, merging and execution, and presents a solution for execution-time merging of multiple clinical guidelines.

Background: Type 2 diabetes mellitus (T2DM) is a major health concern among Navajo Indians. Native Americans and Alaskan Natives (NA/AN) currently have the highest rates of T2DM in the United States (Indian Health Service, 2016). The rate of diabetes on the Navajo Indian reservation is 22% (Partnersinhealth.org, 2009). Major health concerns for patients with T2DM include cardiovascular complications. Treatment is essential to prevent high-risk complications such as, cardiovascular disease (CVD). Purpose: The purpose of this quality improvement project was to implement a clinical decision support tool (CDST) to increase primary care provider awareness of current American Diabetes Association (ADA) statin therapy guidelines. The first objective was to increase the prescription rates of statin medications by 10%. The second objective of this project was to increase the performance target rate by 10%. Setting: This project was implemented at the Gallup Indian Medical Center (GIMC) Family Medicine Clinic. GIMC is located in Gallup, New Mexico. Participants: Participants included primary care providers, six Medical Doctors, two Nurse Practitioners, and one Physician Assistant. Methods: An evidence based clinical support decision tool (CDST) was generated the ADA statin therapy guidelines. Participants were educated on these practice guidelines and the CDST. The CDST was implemented into the electronic health record (EHR) over a four-week period. The provider used the CDST as a point-of-care guide when prescribing statin therapy to those with T2DM. Results: There was a 0.5% increase in the GPRA performance rating at GIMC as well as a 10% increase in prescribed statin therapy medications. There were 253 newly prescribed statin medications during data collection. Conclusion: While this project did not result in significant improvement of statin therapy GPRA performance ratings, a new EHR tool that providers can use to improve patient care was implemented. One outcome was met, there was a 10% increase in statin medication prescriptions. Further studies and future PDSA cycles will be required for testing the effectiveness of CDSTs.

This thesis presents the development and evaluation of a computerized physician-parent decision support tool for a neonatal intensive care unit (NICU), known as Physician and Parent Decision Support (PPADS). The NICU is a specialized hospital unit that treats very-ill neonates. Many difficult care decisions are made daily for this vulnerable population. The PPADS tool aims to augment current NICU decision-making by helping parents and physicians make more informed decisions, improving physician-parent communication, increasing parent decision-making satisfaction, decreasing conflict, and increasing decision efficiency. The development of the PPADS tool followed a five-step methodology: assessing the clinical environment, establishing the design criteria, developing the system design, implementing the system, and performing usability testing. Usability testing of the PPADS tool was performed on neonatologists and on parents of neonates who have graduated (survived) from a tertiary level NICU. The usability testing demonstrated the usefulness and ease of use of the tool.

Around ten percent of patients admitted to National Health Service (NHS) hospitals have experienced a patient safety incident, and an important reason for the high rate of patient safety incidents is medical errors. Research shows that appropriate increase in the use of clinical decision support systems (CDSSs) could help to reduce medical errors and result in substantial improvement in patient safety. However several barriers continue to impede the effective implementation of CDSSs in clinical settings, among which representation of and reasoning about medical knowledge particularly under uncertainty are areas that require refined methodologies and techniques. Particularly, the knowledge base in a CDSS needs to be updated automatically based on accumulated clinical cases to provide evidence-based clinical decision support. In the research, we employed the recently developed belief Rule-base Inference Methodology using the Evidential Reasoning approach (RIMER) for design and development of an online belief rule-based group CDSS prototype. In the system, belief rule base (BRB) was used to model uncertain clinical domain knowledge, the evidential reasoning (ER) approach was employed to build inference engine, a BRB training module was developed for learning the BRB through accumulated clinical cases, and an online discussion forum together with an ER-based group preferences aggregation tool were developed for providing online clinical group decision support.We used a set of simulated patients in cardiac chest pain provided by our research collaborators in Manchester Royal Infirmary to validate the developed online belief rule-based CDSS prototype. The results show that the prototype can provide reliable diagnosis recommendations and the diagnostic performance of the system can be improved significantly after training BRB using accumulated clinical cases.

Machine learning methods provide an opportunity to improve the classification of skin lesions and the early diagnosis of melanoma by providing decision support for general practitioners. So far most studies have been looking at the creation of features that best indicate melanoma. Representation learning methods such as neural networks have outperformed hand-crafted features in many areas. This work aims to evaluate the performance of convolutional neural networks in relation to earlier machine learning algorithms and expert diagnosis. In this work, convolutional neural networks were trained on datasets of dermoscopy images using weights initialized from a random distribution, a network trained on the ImageNet dataset and a network trained on Dermnet, a skin disease atlas.  The ensemble sum prediction of the networks achieved an accuracy of 89.3% with a sensitivity of 77.1% and a specificity of 93.0% when based on the weights learned from the ImageNet dataset and the Dermnet skin disease atlas and trained on non-polarized light dermoscopy images.  The results from the different networks trained on little or no prior data confirms the idea that certain features are transferable between different data. Similar classification accuracies to that of the highest scoring network are achieved by expert dermatologists and slightly higher results are achieved by referenced hand-crafted classifiers.  The trained networks are found to be comparable to practicing dermatologists and state-of-the-art machine learning methods in binary classification accuracy, benign – melanoma, with only little pre-processing and tuning.

Lung cancer affects many individuals each year and accounts for many deaths around the globe. Lung cancer screening is a preventative health measure that has the ability to detect lung cancer earlier. The purpose of this project was to focus on the education of nursing staff within a community health system, with subsequent implementation of an electronic health record clinical decision support system, to create a direct referral pathway to lung cancer screening, delivered through patient education. The concept of prevention was the framework for this project design, which was further organized around the plan-do-study -act model, while taking into consideration the health belief model and theory of interpersonal relations. Using systemized dashboard reports within the electronic health record software, specific variables were targeted for data collection and analyzed for the purpose of this project. Final data demonstrated an increase of triple the programmatic volume of the previous year, directly following the implementation of the above initiative. Further comparative statistics bespeak to the significant needs of the community regarding tobacco dependence and lung cancer screening. High-risk individuals who are current or former smokers will benefit from this initiative by receiving education about lung cancer screening and tobacco dependence treatment while within the care of the community based health system. A nursing-driven pathway to preventative care could also serve other cancer screening programs effectively, as well as be applied to a variety of chronic disease comorbidities to make a significant positive social change.

Indiana University-Purdue University Indianapolis (IUPUI) / Clinical Decision Support is one of the only aspects of health information technology that has demonstrated decreased costs and increased quality in healthcare delivery, yet it is extremely expensive and time-consuming to create, maintain, and localize. Consequently, a majority of health care systems do not utilize it, and even when it is available it is frequently incorrect. Therefore it is important to look beyond traditional guideline-based decision support to more readily available resources in order to bring this technology into widespread use. This study proposes that the wisdom of physicians within a practice is a rich, untapped knowledge source that can be harnessed for this purpose. I hypothesize and demonstrate that this wisdom is reflected by order entry data well enough to partially reconstruct the knowledge behind treatment decisions. Automated reconstruction of such knowledge is used to produce dynamic, situation-specific treatment suggestions, in a similar vein to Amazon.com shopping recommendations. This approach is appealing because: it is local (so it reflects local standards); it fits into workflow more readily than the traditional local-wisdom approach (viz. the curbside consult); and, it is free (the data are already being captured). This work develops several new machine-learning algorithms and novel applications of existing algorithms, focusing on an approach called Bayesian network structure learning. I develop: an approach to produce dynamic, rank-ordered situation-specific treatment menus from treatment data; statistical machinery to evaluate their accuracy using retrospective simulation; a novel algorithm which is an order of magnitude faster than existing algorithms; a principled approach to choosing smaller, more optimal, domain-specific subsystems; and a new method to discover temporal relationships in the data. The result is a comprehensive approach for extracting knowledge from order-entry data to produce situation-specific treatment menus, which is applied to order-entry data at Wishard Hospital in Indianapolis. Retrospective simulations find that, in a large variety of clinical situations, a short menu will contain the clinicians' desired next actions. A prospective survey additionally finds that such menus aid physicians in writing order sets (in completeness and speed). This study demonstrates that clinical knowledge can be successfully extracted from treatment data for decision support.

This critical synopsis of prior work by Luca Parisi is submitted in support of a PhD by Published Work. The work focuses on deriving accurate, reliable and explainable clinical decision support systems as minimum clinically viable solutions leveraging Machine Learning (ML) and evolutionary algorithms, for the first time, to facilitate early diagnostic predictions of Parkinson's Disease and hypothermia in hospitals, as well as prognostic predictions of optimal postoperative recovery area and of chronic hepatitis. Despite the various pathological aetiologies, the underlying capability of ML-based algorithms to serve as a minimum clinically viable solution for predicting early diagnosis and prognosis has been thoroughly demonstrated. Feature selection (FS) is a proven method for increasing the performance of ML-based classifiers for several applications. Although advances in ML, such as Deep Learning (DL), have denied the usefulness of any extrinsic FS by incorporating it in their architectures, e.g., convolutional filters in convolutional neural networks, DL algorithms often lack the required explainability to be understood and interpreted by clinicians within the context of the diagnostic and prognostic tasks of interest. Their relatively complicated architectures, the hardware required for running them and the limited explainability or interpretability of their architectures, the decision-making process – although as assistive tools - driven by the algorithms’ training and predictive outcomes have hindered their application in a clinical setting. Luca Parisi’s work fills this translational research gap by harnessing the explainability of using traditional ML- and evolutionary algorithms-based FS methods for improving the performance of ML-based algorithms and devise minimum viable solutions for diagnostic and prognostic purposes. The work submitted here involves independent research work, including collaborative studies with Marianne Lyne Manaog (MedIntellego®) and Narrendar RaviChandran (University of Auckland). In particular, conciliating his work as a Senior Artificial Intelligence Engineer and volunteering commitment as the President and Research Committee Leader of a student-led association named the “University of Auckland Rehabilitative Technologies Association”, Luca Parisi decided to embark on most research works included in this synopsis to add value to society via accurate, reliable and explainable, hence clinically viable applications of AI. The key findings of these studies are: (i) ML-based FS algorithms are sufficient for devising accurate, reliable and explainable ML-based classifiers for aiding prediction of early diagnosis for Parkinson’s Disease and chronic hepatitis; (ii) evolutionary algorithms-based optimisation is a preferred method for improving the accuracy and reliability of decision support systems aimed at aiding early diagnosis of hypothermia; (iii) evolutionary algorithms-based optimisation methods enable to devise optimised ML-based classifiers for improving postoperative discharge; (iv) whilst ML-based algorithms coupled with ML based FS methods are the minimum clinically viable solution for binary classification problems, ML-based classifiers leveraging evolutionary algorithms for FS yield more accurate and reliable predictions, as reducing the search space and overlapping regions for tackling multi-class classification problems more effectively, which involve a higher number of degrees of freedom. Collectively, these findings suggest that, despite advances in ML, state-of-the-art ML algorithms, coupled with ML-based or evolutionary algorithms for FS, are enough to devise accurate, reliable and explainable decision support systems for performing both an early diagnosis and a prediction of prognosis of various pathologies.

The use of clinical decision support (CDS) systems has grown progressively during the past decades. CDS systems are associated with improved patient safety and outcomes, better prescription and diagnosing practices by clinicians and lower healthcare costs. Quality assurance of these systems is critical, given the potentially severe consequences of any errors. Yet, after several decades of research, there is still no consensual or standardized approach to their verification and validation (V&V). This project is a descriptive and exploratory case study aiming to provide a practical description of how Cambio CDS, a market-leading developer of CDS services, conducts its V&V process. Qualitative methods including semi-structured interviews and coding-based textual data analysis were used to elicit the description of the V&V approaches used by the company. The results showed that the company’s V&V methodology is strongly influenced by the company’s model-driven development approach, a strong focus and leveraging of domain knowledge and good testing practices with a focus on automation and test-driven development. A few suggestions for future directions were discussed.