No abstract

In agile software development, product backlog items (PBI) are used to capture the user requirements prior to the product implementation. Many types of requirements can be observed within a software project. Proper classification of PBI can positively impact the software development process. PBI can be classified into three categories: user stories, foundational stories, and spikes. After the extreme literature survey, no research was held on classifying the PBI into the categories mentioned above. This paper proposed a machine learning (ML) based approach to classify the PBI into three categories. 4,721 PBI were collected from different software projects and manually labelled into the three classes mentioned above. Then the PBI were cleaned using different pre-processing techniques. Classification models were constructed using ML techniques. The performance of each ML model was evaluated using accuracy, precision, recall, and F1 score. Support vector machine (SVM) outperformed other ML models by providing 88% accuracy.

No abstract

BackgroundMany important treatment decisions for patients with rheumatoid arthritis (RA) are conditional on patient preferences and, according to the U.S. National Academy of Medicine, mandate a shared decision making approach (SDM). Furthermore, SDM is being increasingly recognized as an important quality measure. One of the most common preference sensitive decisions in RA is how to escalate care when response to methotrexate monotherapy is inadequate. However, the number of RA medications currently approved makes it challenging for patients to weigh the pros and cons related to each of the treatment options in order to develop a preference.ObjectivesThe objective of this study was to develop representative patient preference phenotypes to enable patient-physician dyads to effectively incorporate patient preferences at the point-of-care.MethodsPeople living with RA were invited to complete a Choice-Based Conjoint analysis survey including seven attributes (route of administration, time to onset of action, bothersome adverse events (AEs), serious AEs, extremely rare AEs, duration of time on the market and affordability) developed iteratively based on patient feedback. Each attribute was described across three or four levels using plain language. Preference phenotypes were identified by applying latent class analysis to the conjoint data. Class solutions were replicated five times from random starting seeds. A five-group solution was chosen based on Akaike's information criterion. We calculated the percentage of importance assigned to each attribute and performed simulations to estimate preferences for triple therapy, SC and IV biologics, or tofacitinib.Results1100 U.S. subjects recruited via the CreakyJoints online patient community completed the survey. Of these, 49 were eliminated because they completed the survey in less than 10 minutes and an additional 45 people were excluded because they did not respond correctly to a dominant choice task. The mean age was 51.7 (11.2). The majority were female; (92%) and Caucasian (93%). Preferences (assuming low cost across options), and the reasons underlying each respondent's preference, clustered into five groups (Figure 1). There were no differences in the distribution of demographic or clinical characteristics across the five groups. Phenotypes were created based on the stated preference data.ConclusionsRA patients' preferences vary and can be classified into distinct phenotypes. Ongoing research is evaluating whether enabling patients to identify with a preference phenotype facilitates SDM at the point-of-care.AcknowledgementsThis research was supported by a grant from the Rheumatology Research Foundation.Disclosure of InterestNone declared