Semi-automated Code Research Articles

Breathing new life into death certificates: Extracting handwritten cause of death in the LIFE-M project

The demographic and epidemiological transitions of the past 200 years are well documented at an aggregate level. Understanding differences in individual and group risks for mortality during these transitions requires linkage between demographic data and detailed individual cause of death information. This paper describes the digitization of almost 185,000 causes of death for Ohio to supplement demographic information in the Longitudinal, Intergenerational Family Electronic Micro-database (LIFE-M). To extract causes of death, our methodology combines handwriting recognition, extensive data cleaning algorithms, and the semi-automated classification of causes of death into International Classification of Diseases (ICD) codes. Our procedures are adaptable to other collections of handwritten data, which require both handwriting recognition and semi-automated coding of the information extracted.

Application of a Machine Learning-Based Decision Support Tool to Improve an Injury Surveillance System Workflow.

Emergency department (ED)-based injury surveillance systems across many countries face resourcing challenges related to manual validation and coding of data. This study describes the evaluation of a machine learning (ML)-based decision support tool (DST) to assist injury surveillance departments in the validation, coding, and use of their data, comparing outcomes in coding time, and accuracy pre- and postimplementations. Manually coded injury surveillance data have been used to develop, train, and iteratively refine a ML-based classifier to enable semiautomated coding of injury narrative data. This paper describes a trial implementation of the ML-based DST in the Queensland Injury Surveillance Unit (QISU) workflow using a major pediatric hospital's ED data comparing outcomes in coding time and pre- and postimplementation accuracies. The study found a 10% reduction in manual coding time after the DST was introduced. The Kappa statistics analysis in both DST-assisted and -unassisted data shows increase in accuracy across three data fields, that is, injury intent (85.4% unassisted vs. 94.5% assisted), external cause (88.8% unassisted vs. 91.8% assisted), and injury factor (89.3% unassisted vs. 92.9% assisted). The classifier was also used to produce a timely report monitoring injury patterns during the novel coronavirus disease 2019 (COVID-19) pandemic. Hence, it has the potential for near real-time surveillance of emerging hazards to inform public health responses. The integration of the DST into the injury surveillance workflow shows benefits as it facilitates timely reporting and acts as a DST in the manual coding process.

Підвищення продуктивності коду мовою РYТНОN з використанням техніки переписувальних правил

Python is a popular programming language used in many areas, but its performance is significantly lower than many compiled languages. We propose an approach to increasing performance of Python code by transforming fragments of code to more efficient languages such as Cython and C++. We use high-level algebraic models and rewriting rules technique for semi-automated code transformation. Performance-critical fragments of code are transformed into a low-level syntax model using Python parser. Then this low-level model is further transformed into a high-level algebraic model that is language-independent and easier to work with. The transformation is automated using rewriting rules implemented in Termware system. We also improve the constructed high-level model by deducing additional information such as data types and constraints. From this enhanced high-level model of code we generate equivalent fragments of code using code generators for Cython and C++ languages. Cython code is seamlessly integrated with Python code, and for C++ code we generate a small utility file in Cython that also integrates this code with Python. This way, the bulk of program code can stay in Python and benefit from its facilities, but performance-critical fragments of code are transformed into more efficient equivalents, improving the performance of resulting program. Comparison of execution times between initial version of Python code, different versions of transformed code and using automatic tools such as Cython compiler and PyPy demonstrates the benefits of our approach – we have achieved performance gains of over 50x compared to the initial version written in Python, and over 2x compared to the best automatic tool we have tested.

Protesting Parties in Europe: A comparative analysis

The article provides the first large-scale study of protest activities by political parties. The empirical analysis draws on original protest event data for 30 European countries based on semi-automated coding of news agencies. The article innovates by (a) proposing a standardized indicator for the extent to which protest and electoral politics relate to each other, (b) showing that parties’ involvement in protests differs across political contexts, and (c) mapping the profile of a typical party-sponsored event and a typical protesting party. Despite long-term trends toward differentiated modes of interest intermediation, the results indicate that a wide range of parties does protest. However, in highly differentiated contexts, the typical protesting party mirrors the outsider image of movement parties: it does not belong to a mainstream party family and has no government experience. By contrast, more strategic factors, such as opposition status, drive parties to the streets in less differentiated contexts.

Protest Event Analysis as a Tool for Political Mobilization Studies

In this paper, we analyze the scope and limits of protest event-analysis as a tool for studying political mobilization. Event-analysis is a form of content-analysis and allows for the reconstruction of the dynamics of contention and its key characteristics on the basis of different text sources (police and media reports, participatory observations, etc.). The inception of protest event-analysis is closely linked to the development of the comparative method in the social sciences and to the competition between the theories of collective action and social movements. The demand for wide cross-national comparisons and the quantification of social-political phenomena in addition to the pioneering works of Charles Tilly created the preconditions for a systematic collection of information on protest events. Four generations of event analysis led to the formulation of key concepts of collective action theory (“political opportunity structure”, “cycle of protest”, “repertoire of contention”, etc.), and also improved the techniques and procedures of data collection process, for example, by introducing the triangulation of sources using semi-automated coding. We also analyze the sources of systematic bias in data collection process (selective, descriptive, and research) and the possible means of correction. In addition, we compare the event catalogs of Russian protests and point to specifics of this case, namely, spatial heterogeneity, poor quality of media reports in absence of alternative sources, and the bias in media coverage towards big cities. Using the experience of the “Contention Politics in Russia” database development, we propose systematic methods and means which address the challenges of protest event-analysis in regards to Russia.

A practical tool for public health surveillance: Semi-automated coding of short injury narratives from large administrative databases using Naïve Bayes algorithms

Public health surveillance programs in the U.S. are undergoing landmark changes with the availability of electronic health records and advancements in information technology. Injury narratives gathered from hospital records, workers compensation claims or national surveys can be very useful for identifying antecedents to injury or emerging risks. However, classifying narratives manually can become prohibitive for large datasets.The purpose of this study was to develop a human–machine system that could be relatively easily tailored to routinely and accurately classify injury narratives from large administrative databases such as workers compensation. We used a semi-automated approach based on two Naïve Bayesian algorithms to classify 15,000 workers compensation narratives into two-digit Bureau of Labor Statistics (BLS) event (leading to injury) codes. Narratives were filtered out for manual review if the algorithms disagreed or made weak predictions.This approach resulted in an overall accuracy of 87%, with consistently high positive predictive values across all two-digit BLS event categories including the very small categories (e.g., exposure to noise, needle sticks). The Naïve Bayes algorithms were able to identify and accurately machine code most narratives leaving only 32% (4853) for manual review. This strategy substantially reduces the need for resources compared with manual review alone.

Flexible neuronal network simulation framework using code generation for NVidia® CUDA™

Simulating large scale computer models of brain structures with spiking neuronal networks has become increasingly popular and feasible with the advent of general purpose computing on graphical processing units (GPGPU). Modern graphics cards, such as the NVidia® range supporting the common unified device architecture (CUDA™) provide massively parallel computing architectures for this purpose. Earlier GPU implementations of neural networks, including my own earlier work [1,2], were customized for specific models, and optimized and tested with specific hardware. Recently, more general spiking neuronal network simulators have been developed [3,4] that allow the definition of the network connectivity and neuron- and synapse parameters at runtime. However, the simulators are still quite specific in using a single neuron model (typically Izhikevich neurons), synapse model (typically stateless synapses with delay) and have been tested for a typical model type and on specific GPU hardware. In this work I present a framework of semi-automated code generation for simulating neuronal networks on GPU hardware. Using code generation to build a specific simulator engine for each individual network model has important advantages: (i) The simulator system can provide a large choice of different neuron and synapse models for use in simulations without creating any overheads or performance losses in the actual simulation code. It also allows the inclusion of user-defined models without the necessity for a user to understand the GPU code. (ii) The generated simulator software can be optimized for the available GPU hardware and for the structure of the specific model. (iii) The framework is intrinsically extensible: New GPU optimization strategies can be added and strategies of existing simulators can be included in the generated code for situations where they are effective. An embryonic beta version of such a framework has been built and optimized for simulating neuronal networks with an anatomical structure (separate neuron populations that are densely connected), building on our earlier work on neuronal network models [1,2] of the olfactory system of insects [5,6]. The prototype framework consists of a C++ source library that generates CUDA kernels and runtime code according to a user-specified neuronal network model. Unlike existing simulators [5,6], it exploits that most parameters of neuronal network simulations are known at compile time and do not change during a simulation. These parameters are hard-coded into the CUDA kernels saving valuable register and shared memory space. In practice, the system is used by defining a neuronal network in a C++ class and compiling and executing the code generation software. The generated C++/CUDA code is compiled with user-side simulation code into a lean stand-alone executable. I have tested the system on an NVidia Quadro FX 5800 device hosted in a PC with Intel Xeon quad core CPU with 8MB cache and 12 GB RAM. I have observed variable GPU versus CPU speedups between none and 76x. The peak observed spike delivery per second was 2.4 billion and the largest simulated network had about 1 million neurons and 635 million synapses (limited by the 4GB device memory on the FX 5800). In the future I will extend the system with additional model elements, optimizations and improved API for use by the CNS community.

A REPOSITORY FOR COMPONENT-BASED EMBEDDED SOFTWARE DEVELOPMENT

The rapid growth in the demand for embedded systems and the increased complexity of embedded software pose an urgent need for advanced embedded software development techniques. Software technology is shifting toward semi-automated code generation and integration of systems from components. Component-based development (CBD) techniques can significantly reduce the time and cost for developing software systems. Furthermore, effective component retrieval is a fundamental issue in CBD. In this paper, we address the issues in designing software repositories for embedded software components. We develop an On-line Repository for Embedded Software (ORES) to facilitate component management and retrieval. ORES uses an ontology-based approach to facilitate repository browsing and effective search. To allow easy browsing of ORES, we analyze the typical ontology relations for software components and develop a Merging and Echoing technique to convert the ontology into a hierarchy suitable for browsing, but without the loss of any critical semantic information contained in the ontology. We also develop an algorithm for grouping search results based on the ontology. Thus, we can display search result groups to avoid having to display a large number of search results or having to prune the results and risk reducing the recall factor. Another important aspect in embedded software is the set of nonfunctional requirements and properties. In ORES, we develop an XML-based specification method to capture nonfunctional properties as well as functional characteristics of components and enable retrieval of relevant components based on these specifications.

Efficiency of a Semiautomated Coding and Review Process for Notification of Critical Findings in Diagnostic Imaging

When a significant unexpected finding such as malignancy is noted on a study, the standard of care generally holds that the radiologist communicate the findings to the referring physician and document the communication in the radiology report. Despite this standard, for a variety of reasons it remains possible that the direct care provider might receive such notification but not initiate an appropriate workup. On the basis of prior root cause analysis, we developed and instituted a semiautomated process for notification of critical diagnostic imaging findings. We now report our 12-month experience with the process. A diagnostic code was attached to every radiology report. When a significant unexpected finding occurred, our radiologists, in addition to contacting the appropriate clinician, gave the report the designation code 8. On a weekly basis, a list of code 8 cases was passed to the cancer registrar at our institution, who tracked the cases to ensure that they were appropriately followed up. In the 12-month period after initiation of this system, we performed 37,736 radiologic examinations at our institute. Of these, 395 cases were given code 8. All code 8 cases were followed up by the tumor registrar. In 35 cases, no workup was documented after 2 weeks. Of these, eight cases would have been completely lost to follow-up if this safety net had not been in place. Failures of communication, documentation errors, and various system failures may lead to an untoward outcome for the patient. We devised a simple system to ensure that significant unexpected findings on imaging received appropriate attention. An additional level of redundancy has increased the probability of optimal patient outcome.

Simplifying Qualitative Data Analysis Using General Purpose Software Tools

This article shows how clever but simple use of word-processing functions can provide many features of special-purpose software designed for analyzing text. For many qualitative research projects, and for students who are learning computerassisted analysis of text, the Microsoft Word functions outlined here may be all that are required. Examples are given showing how Microsoft Word can be used for coding and retrieving, semiautomated coding and inspection, creating hierarchies of code categories via indexing, global editing of theme codes, coding of “face-sheet” data, exploring relationships between face-sheet codes and conceptual codes, quantifying the frequency of code instances, and annotating text. The techniques outlined can be used for analyzing and managing many kinds of data, including key informant interviews, focus groups, document and literature reviews, and open-ended survey questions.