Open Coding Research Articles

A Data-Driven Approach to Inguinal Hernia Repairs in Infants and Children.

Introduction: Laparoscopic inguinal hernia repair has become increasingly popular in children. The laparoscopic technique inherently assesses the contralateral processus vaginalis, reducing the risk of metachronous contralateral hernias. We hypothesized that primary laparoscopic repair would be associated with lower rates of subsequent hernia repair in the youngest patients, in whom metachronous contralateral hernias are most common. Materials and Methods: The Pediatric Health Information System database was queried for patients 0-15 years old, who underwent inguinal hernia repair between 2016 and 2022. The primary outcome was the need for subsequent hernia repair. Current Procedural Terminology (CPT) and ICD-10 procedure codes were used to determine laparoscopic versus open repair. Patients were excluded if the only recorded code was for recurrent hernia or if both laparoscopic and open codes were present for the same procedure. Results: A total of 109,456 patients were included in the study, with 20,338 patients (18.58%), undergoing laparoscopic inguinal hernia repair initially, and 2535 patients (2.32%) requiring a second hernia repair. Patients 6 months old and younger undergoing unilateral laparoscopic repair were less likely to require subsequent surgery (OR 0.82, 95% CI = 0.69-0.96). Across all ages, open bilateral repair less often required subsequent repairs (OR 1.93, 95% CI 1.48-2.51). Conclusion: Laparoscopic unilateral inguinal hernia repair decreases the need for subsequent surgical repair in infants 6 months and younger. No difference was detected in older patients. Open repair of bilateral hernias decreases the need for a second hernia operation in all age groups, suggesting that open repair is more durable.

Development of an MRI Radiomic Machine-Learning Model to Predict Triple-Negative Breast Cancer Based on Fibroglandular Tissue of the Contralateral Unaffected Breast in Breast Cancer Patients

Aim: The purpose of this study was to develop a radiomic-based machine-learning model to predict triple-negative breast cancer (TNBC) based on the contralateral unaffected breast’s fibroglandular tissue (FGT) in breast cancer patients. Materials and methods: This study retrospectively included 541 patients (mean age, 51 years; range, 26–82) who underwent a screening breast MRI between November 2016 and September 2018 and who were subsequently diagnosed with biopsy-confirmed, treatment-naïve breast cancer. Patients were divided into training (n = 250) and validation (n = 291) sets. In the training set, 132 radiomic features were extracted using the opensource CERR platform. Following feature selection, the final prediction model was created, based on a support vector machine with a polynomial kernel of order 2. Results: In the validation set, the final prediction model, which included four radiomic features, achieved an F1 score of 0.66, an area under the curve of 0.71, a sensitivity of 54% [47–60%], a specificity of 74% [65–84%], a positive predictive value of 84% [78–90%], and a negative predictive value of 39% [31–47%]. Conclusions: TNBC can be predicted based on radiomic features extracted from the FGT of the contralateral unaffected breast of patients, suggesting the potential for risk prediction specific to TNBC.

Design of unmanned aerial vehicle formation keeping controller based on improved consistency algorithm design

This article conducts in-depth research on the problems of large gain convergence oscillation and simultaneous application of translational rotation in traditional consistency algorithms. First, based on the traditional consistent formation algorithm, a virtual long machine is introduced, and a new norm is defined to extend the original consistent algorithm to solve the problem of large gain convergence oscillation in the Linearizability algorithm. Secondly, in response to the issue of the inability to maintain the formation caused by the application of translational rotation in traditional consistent formation processes, the maximum distance is introduced as feedback to the reference controller to modify the reference input trajectory to ensure good tracking of the formation during translational rotation. Finally, a joint semi physical simulation was conducted using the open-source flight control Ardupilot and the open-source simulation platform Airsim to further validate the feasibility of the improved algorithm. The results of this study indicate that the proposed algorithm has significant improvements in solving the problems existing in traditional consistency protocols and has good feasibility in practical applications.

Open-source software integration: A tutorial on species distribution mapping and ecological niche modelling

Over the last decade, access to global data has become increasingly critical for research, allowing insights into diverse biological, environmental and societal questions at a macro scale. Digitisation has greatly enhanced the use of herbarium data in the analysis of species distributions and ecological niche modelling. Yet, sources on modelling and mapping methodology using open-source software is greatly lacking for beginners. We have created a replicable and thorough tutorial to visualise species occurrence data and exploratory analysis that was developed by undergraduates with broad backgrounds and levels of experience. This tutorial integrates the open-source programmes QGIS, MaxEnt and R to develop distribution maps, using bryophytes as a case study, to promote the accessibility of open-source software and remote access learning. This tutorial has already set the foundation for further research into distribution modelling of rare Illinois bryophytes to better understand the potential impact of climate change.

Folk Models of Loot Boxes in Video Games

Regulations require video games to provide transparency regarding loot box odds to keep players informed, leading many games to disclose probabilities in various ways; yet, the extent of players' comprehension of loot box mechanics remains unclear. We performed a content analysis on 80 online posts to understand players' perceptions of loot box odds in two popular video games (Genshin Impact and Honkai: Star Rail). We then conducted semi-structured interviews with 24 players to explore the causes of these folk models across more games. Utilizing a bottom-up open coding approach, we created a taxonomy of folk models players have about loot boxes. We found that participants generally possessed inaccurate mental models of how loot boxes work, and they wanted game companies to enhance loot box transparency in three areas of probability disclosures: granularity, longitude, and scope.

Land cover and drought risk assessment in Türkiye's mountain regions using neutrosophic decision support system.

Earth observation (EO) provides dynamic scientific methods for tracking and defining ecological parameters in mountainous regions. Open-source platforms are frequently utilized in this context to efficiently collect and evaluate spatial data. In this study, we used Collect Earth (CE), an open-source land monitoring platform, to reveal and assess land cover, land cover change, and relevant ecological parameters such as drought risk. Mountain ecosystems were subject to an evaluation for the first time by combining remote sensing with a hybridization of Decision-Making Trial and Evaluation Laboratory (DEMATEL), analytic hierarchy process (AHP), and Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) for neutrosophic sets in risk assessment problems of several connected criteria. The high and dispersed high alpine environment of Türkiye accommodates land with relatively less human influence, making it suitable to observe climate change impacts. In the framework of the study, we evaluated more than two decades (2000-2022) of land use and land cover (LULC) changes in the mountain regions of the country. Using nine identified ecological parameters, we also evaluated drought risk. The parameters included were the LULC classes and their change, elevation, slope, aspect, precipitation, temperature, normalized difference vegetation index (NDVI), water deficit, and evapotranspiration (ET). The risk map we produced revealed a high to very high drought risk for almost throughout the Türkiye's mountainous areas. We concluded that integrating geospatial techniques with hybridization is promising for mapping drought risk, helping policymakers prepare effective drought mitigation measures to reasonably adapt to climate change impacts.

Equity in Literacy Education: A Case Study of Tier 2 Personalized Instruction

According to the NAEP, proficiency in reading is reported to be 67% in 4th-grade students. General education classroom teachers are responsible for providing Tier 1 and 2 instruction, including interventions in Tier 2. However, many teachers are unsure about implementing interventions and have had negative experiences with Response to Intervention (RTI), finding the process frustrating and overwhelming. This case study examines the effectiveness of teachers who have seen an increase in reading levels and student interest. The research explores the teachers’ perceptions, experiences, and skills of the current implementation of RTI using Gibbs’ Reflective Practice as a theoretical framework. Data collected from individual interviews and observations are analyzed using open coding. Themes and concepts are identified from the interview and observational data. These findings inform educators and higher education teacher educators about Tier 2 implementation and equitable, personalized instruction in the RTI process, aiming to strengthen the implementation and effectiveness of Tier 2 interventions in the general education classroom.

Analytical model for RC coupling beam considering axial restraint

A novel three-dimensional distributed plasticity model element is formulated and implemented into an open-source computational platform (OpenSees) to enhance nonlinear modeling approaches for lateral force resisting systems utilizing diagonally reinforced concrete coupling beams. The model incorporates a fiber-based concrete cross-section along with truss elements to represent the diagonal reinforcement and contact elements to model reinforcing bar slip at the beam-wall interface. The model is validated using test results from five different experiments that include variations in cross-section geometries, reinforcement configurations, and boundary conditions. Notably, two test specimens shared identical attributes except for their cross-sectional shapes (with and without a slab), whereas another pair were identical except for the degree of axial restraint provided. The final specimen utilized both longitudinal and diagonal reinforcement along the length of the beam. The model accurately predicted overall shear versus chord rotation responses with errors < 5 % and axial growth responses with errors < 30 %. The proposed model provides enhanced modeling options for nonlinear analysis by enabling sensitivity studies to address the potential impacts of coupling beam axial restraint on coupling beam and coupled wall behavior.

TestCompareR: an R package to compare two binary diagnostic tests using paired data

Background Binary diagnostic tests are commonly used in medicine to answer a question about a patient’s clinical status, most commonly, do they or do they not have some disease. Recent advances in statistical methodologies for performing inferential statistics to compare commonly used test metrics for two diagnostic tests have not yet been implemented in a robust statistical package. Methods Up-to-date statistical methods to compare the test metrics achieved by two binary diagnostic tests are implemented in the new R package testCompareR. The output and efficiency of testCompareR is compared to the only other available package which performs this function, DTComPair, as well as an open-source program, compbdt, using a motivating example. Results testCompareR achieves similar results to DTComPair using statistical methods with improved coverage and asymptotic performance. Further, testCompareR is faster than the currently available package and requires fewer pre-processing steps in order to produce accurate results. Conclusions testCompareR provides a new tool to compare the test metrics for two binary diagnostic tests compared with the gold standard. This tool allows flexible inputs, which minimises the need for data pre-processing, and operates in very few steps, so that it is easy to use even for those less experienced with R. testCompareR achieves results comparable to those computed by DTComPair, using optimised statistical methods and with improved computational efficiency.

Impact-based forecasting of tropical cyclone-related human displacement to support anticipatory action

Tropical cyclones (TCs) displace millions every year. While TCs pose hardships and threaten lives, their negative impacts can be reduced by anticipatory actions like evacuation and humanitarian aid coordination. In addition to weather forecasts, impact forecast enables more effective response by providing richer information on the numbers and locations of people at risk of displacement. We introduce a fully open-source implementation of a globally consistent and regionally calibrated TC-related displacement forecast at low computational costs, combining meteorological forecast with population exposure and respective vulnerability. We present a case study of TC Yasa which hit Fiji in December 2020. We emphasise the importance of considering the uncertainties associated with hazard, exposure, and vulnerability in a global uncertainty analysis, which reveals a considerable spread of possible outcomes. Additionally, we perform a sensitivity analysis on all recorded TC displacement events from 2017 to 2020 to understand how the forecast outcomes depend on these uncertain inputs. Our findings suggest that for longer forecast lead times, decision-making should focus more on meteorological uncertainty, while greater emphasis should be placed on the vulnerability of the local community shortly before TC landfall. Our open-source codes and implementations are readily transferable to other users, hazards, and impact types.

Exploring microplastic distribution in Western North American snow

Microplastic (MP) transport in the atmosphere, one of the least studied environmental compartments because of the relatively small size of air-borne MPs and the challenges in identifying them, may be inferred from their occurrence in snowfall. In this study, 11 sites across western coastal North America were sampled and analyzed for MP presence in fresh snowfall, months-old summer surface snow, and stratified deposits in snow pits. MPs were detected and characterized using a method integrating linear array µ-Fourier Transform Spectroscopy (µFTIR) and batch spectral analysis with open-source platform Open Specy. Recovery rate analysis from sample filtration to data analysis was conducted, and analysis of field or laboratory blanks suggested negligible contamination (≤ 1 polyamide fragment per blank). Concentrations of MPs in the fresh snowfall of remote sites and those proximal to sources were 5.1–150.8 p/L and 104.5–325 p/L of snowmelt water, respectively. Summer surface snow that was several months old had MP concentrations ranging from 57.5–539 p/L of meltwater, and snow sampled at different depths within a snowpack had concentrations ranging from 35–914 p/L. Our results demonstrate a streamlined method that may be used for measuring MPs in remote or pristine environments, contributing to a better understanding of long-range MP transport.

Gaspery: Optimized Scheduling of Radial Velocity Follow-up Observations for Active Host Stars

Radial velocity (RV) follow-up is a critical complement of transiting exoplanet surveys like the Transiting Exoplanet Survey Satellite, both for validating discoveries of exoplanets and measuring their masses. Stellar activity introduces challenges to interpreting these measurements because the noise from the host star, which is often correlated in time, can result in high RV uncertainty. A robust understanding of stellar activity and how its timescales interact with the observing cadence can optimize limited RV resources. For this reason, in the era of oversubscribed, high-precision RV measurements, folding stellar activity timescales into the scheduling of observation campaigns is ideal. We present gaspery, an open-source code implementation to enable the optimization of RV observing strategies. Gaspery employs a generalized formulation of the Fisher information for RV time series, which also incorporates information about stellar correlated noise. We show that the information contained in an observing strategy can be significantly affected by beat frequencies between the orbital period of the planet, the stellar rotation period, and the observation epochs. We investigate how the follow-up observing strategy will affect the resulting RV uncertainty, as a function of stellar properties such as the spot decay timescale and rotation period. We then describe two example use cases for gaspery: (1) calculating the minimum number of observations to reach an uncertainty tolerance in a correlated noise regime and (2) finding an optimal strategy given a fixed observing budget. Finally, we outline a prescription for selecting an observing strategy that is generalizable to different targets.

Adapting OGC's SensorThings API and data model to support data management and sharing for environmental sensors

Software is critical in managing environmental sensor data. The Open Geospatial Consortium (OGC) developed the “OGC SensorThings API” (STA) standard to address variability across sensors, observed variables, platforms, and protocols, facilitating development of sensing and Internet of Things applications. This paper details a Python/Django implementation of the STA application programming interface (API) and a PostgreSQL/Timescale implementation of the STA data model, enhancing availability of robust software for management and sharing of environmental sensor data. STA offers a RESTful interface with JSON data encoding, aligning with modern development patterns and facilitating interoperability. Integration of metadata from the Observations Data Model ensures data can be adequately described and interpreted. STA's flexibility allows lightweight query responses or comprehensive metadata inclusion, and a complementary data management API enhances use of STA within multi-user systems. Open-source code and deployment instructions in GitHub enable standalone or cloud deployments, enhancing accessibility and usability for researchers and practitioners.

Emergent ecology in a microscale model of the surface ocean.

Microbial processes operate at the microscale, which is not resolved by existing ecosystem models. Here, we present a novel model that simulates a 1 mL three-dimensional cube using a hybrid Lagrangian-Eulerian approach, at ecologically relevant timescales. The model simulates individual microbes, including three phytoplankton size classes with healthy, senescent, and dead lifecycle stages; copiotrophic and oligotrophic heterotrophic bacteria; and dissolved organic matter at 50 µm resolution. Diffusion, shear, sedimentation, chemotaxis, and attachment processes are explicitly resolved. The emerging quantitative representation of the ecosystem shows that (1) copiotrophs grow mostly attached to eukaryotic phytoplankters and get almost all of their carbon from them vs. oligotrophs that grow on exudates and lysates of cyanobacteria; (2) contrasting diel patterns in substrate appearance in the phycosphere vs. ambient water and growth of particle-associated copiotrophs vs. free-living oligotrophs; (3) attached bacteria reduce carbon flux from the phycosphere, lowering chemotactic efficiency toward eukaryotes below that toward cyanobacteria; (4) shear reduces chemotactic efficiency and fitness of the copiotroph; and (5) the main benefit of chemotaxis is to locate attachment partners. These patterns are consistent with available observations. Our study provides insights into the microscale ecology of marine bacteria, and the open-source code is a tool for further research in this area.IMPORTANCEA large amount of global CO2 fixation is performed by marine phytoplankton, and a substantial fraction of that is released as dissolved organic carbon and further processed by heterotrophic bacteria. The interaction between phytoplankton and bacteria, i.e., the carbon flux between them, is therefore an important process in the global carbon and climate system. Some bacteria have developed specialized behavioral traits, like swimming and attachment, to increase their carbon acquisition. These interactions occur at the micrometer scale, for example, the immediate vicinity of phytoplankters (the phycosphere), but existing biogeochemical models typically only simulate down to the 1 meter vertical or ~100 kilometer horizontal scale. We present a new microscale model and use it to predict fluxes and other features in the surface ocean. The model makes important predictions about the fluxes between various types of phytoplankton and bacteria and the role of behavioral traits, and it provides a basis and tool for further research in this area.

Modular Microgrid Technology with a Single Development Environment Per Life Cycle

The life cycle of a microgrid covers all the stages from idea to implementation, through exploitation until the end of its life, with a lifespan of around 25 years. Covering them usually requires several software tools, which can make the integration of results from different stages difficult and may imply costs being hard to estimate from the beginning of a project. This paper proposes a unified platform composed of four modules developed in MATLAB 2022b, designed to assist all the processes a microgrid passes through during its lifetime. This entire platform can be used by a user with low IT knowledge, because it is completed with fill-in-the-blank alone, as a major advantage. The authors detail the architecture, functions and development of the platform, either by highlighting the novel integration of existing MATLAB tools or by developing new ones and designing new user interfaces linked with scripts based on its complex mathematical libraries. By consolidating processes into a single platform, the proposed solution enhances integration, reduces complexity and provides better cost predictability throughout the project’s duration. A proof-of-concept for this platform was presented by applying the life-cycle assessment process on a real-case study, a microgrid consisting of a photovoltaic plant, and an office building as the consumer and energy storage units. This platform has also been developed by involving students within summer internships, as a process strengthening the cooperation between industry and academia. Being an open-source application, the platform will be used within the educational process, where the students will have the possibility to add functionalities, improve the graphical representation, create new reports, etc.

Testing and results of an open-source radiation epidemiology model using the Goiânia accident

In the event of dispersed radioactive materials, whether from accidental orphan sources or deliberate use of radiological dispersal devices (RDD) or radiological exposure devices (RED), free open-source modelling codes can greatly assist in forecasting the dispersion of the radiation following the event. Several codes are currently available to quickly calculate the progression of radiological dispersion. However, most of these codes can only simulate the evolution of the threat for limited times after the event and over relatively short distances from the location. In order to predict the transport of radioactive material over long distances and for long times, and thus prevent its expected effects on the exposed population, specific epidemiological codes can be used, taking into account the characteristic of the radiation. If it is considered that radioactive material can be deposited on unsuspecting people who continue their daily activities after exposure, it can be assumed that these people unintentionally carry this radioactive material over long distances. This scenario is comparable to viral vectors of a hypothetical virus designed to mimic the physical characteristics of radiation. In this work, the free open-source spatio-temporal epidemiological modeller (STEM) tool is used to simulate the spread of a chimeric viral agent with specific characteristics of Ebola and COVID-19, designed to replicate the biological conditions caused by exposure to a Cs-137 source for an individual unaware of the risk. The goal is to predict the territorial spread of radioactive material caused by a CBRNe event, such as orphan sources or the use of a RDD or a RED, and its possible effects on the affected population. This supports decision-makers in forecasting the consequences of radioactive material spread and thus helps in reducing the risk. The code was tested comparing its results with the real case of the famous 1987 Goiânia radiological accident. The results show that the developed code was indeed able to accurately represent the number of contaminated individuals and the number of casualties within a month of the initial exposure, with a distribution of radioactive material in the territory similar to that actually caused by the Goiânia accident.

Nurses and policymakers role in preparing adolescents with HIV for self-disclosure in Eswatini.

Nurses in Eswatini are best positioned to assist adolescents living with HIV to disclose their status to others. Nonetheless, it is evident that many nurses are not actively involved in the disclosure process. The aim of this study was to explore the process of nurses in preparing adolescents for self-disclosure and describe the role of policymakers in enabling adolescents to disclose their HIV status to others. The study was conducted in four facilities, one from each of the four regions of the country, after getting ethical clearance from a Public University Higher Degrees Ethics Committee and the Eswatini Health and Human Research Review Board. In-depth interviews were conducted on 28 participants: 24 nurses and 4 policymakers. The three steps of open, axial and selective coding were used to analyse data until theoretical saturation was achieved. Adolescents were assisted to disclose by providing them with HIV information to empower them, encouraged to enrol in teen club because it created a conducive environment for peer-to-peer support, and they were given ongoing psychosocial support to prepare them for self-disclosure. Adolescent HIV management workshops were not routinely done because such training relied on funders. Nurses are not preparing adolescents satisfactorily to disclose. Prioritising the training of nurses would lead to a remarkable increase in the rate of HIV self-disclosure by Swati adolescents.Contribution:This study is the first of its kind in Eswatini, and the results will contribute to the review of HIV management guidelines and promote adolescent self-disclosure.

FreeTO - Freeform 3D topology optimization using a structured mesh with smooth boundaries in Matlab

Topology optimization has revolutionized the design of structures for various applications, particularly with the advancement of additive manufacturing. However, existing open-source codes for topology optimization have limitations, such as restricted domain initialization and lack of a CAD output after optimization. A novel open-source Matlab code, FreeTO, is presented, and it addresses these limitations by enabling the initialization of 3D arbitrary geometries and providing an STL file post-optimization. FreeTO utilizes a structured mesh and a smooth-edge (boundary) algorithm to generate smooth topological boundaries. The code is demonstrated through six practical design cases, showcasing its effectiveness in compliance minimization, compliant mechanisms, and self-supporting problems. FreeTO offers a user-friendly, all-in-one topology optimization package, making it an invaluable tool for educators, researchers, and practitioners. Future developments will focus on eliminating a few geometrical deviations in the optimized topologies, incorporating speedups, and extending the code to apply to more applications.

Knowledge Transfer from High-Resource to Low-Resource Programming Languages for Code LLMs

Over the past few years, Large Language Models of Code (Code LLMs) have started to have a significant impact on programming practice. Code LLMs are also emerging as building blocks for research in programming languages and software engineering. However, the quality of code produced by a Code LLM varies significantly by programming language. Code LLMs produce impressive results on high-resource programming languages that are well represented in their training data (e.g., Java, Python, or JavaScript), but struggle with low-resource languages that have limited training data available (e.g., OCaml, Racket, and several others). This paper presents an effective approach for boosting the performance of Code LLMs on low-resource languages using semi-synthetic data. Our approach, called MultiPL-T, generates high-quality datasets for low-resource languages, which can then be used to fine-tune any pretrained Code LLM. MultiPL-T translates training data from high-resource languages into training data for low-resource languages in the following way. 1) We use a Code LLM to synthesize unit tests for commented code from a high-resource source language, filtering out faulty tests and code with low test coverage. 2) We use a Code LLM to translate the code from the high-resource source language to a target low-resource language. This gives us a corpus of candidate training data in the target language, but many of these translations are wrong. 3) We use a lightweight compiler to compile the test cases generated in (1) from the source language to the target language, which allows us to filter our obviously wrong translations. The result is a training corpus in the target low-resource language where all items have been validated with test cases. We apply this approach to generate tens of thousands of new, validated training items for five low-resource languages: Julia, Lua, OCaml, R, and Racket, using Python as the source high-resource language. Furthermore, we use an open Code LLM (StarCoderBase) with open training data (The Stack), which allows us to decontaminate benchmarks, train models without violating licenses, and run experiments that could not otherwise be done. Using datasets generated with MultiPL-T, we present fine-tuned versions of StarCoderBase and Code Llama for Julia, Lua, OCaml, R, and Racket that outperform other fine-tunes of these base models on the natural language to code task. We also present Racket fine-tunes for two very recent models, DeepSeek Coder and StarCoder2, to show that MultiPL-T continues to outperform other fine-tuning approaches for low-resource languages. The MultiPL-T approach is easy to apply to new languages, and is significantly more efficient and effective than alternatives such as training longer.

C anton MT: Investigating Back-Translation and Model-Switch Mechanisms for Cantonese-English Neural Machine Translation

This paper investigates the development and evaluation of machine translation models from Cantonese to English (and backward), where we propose a novel approach to tackle low-resource language translations. Despite recent improvements in Neural Machine Translation (NMT) models with Transformer-based architectures, Cantonese, a language with over 80 million native speakers, has below-par State-of-the-art commercial translation models due to a lack of resources. The main objectives of the study are to develop a model that can effectively translate Cantonese to English and evaluate it against state-of-the-art commercial models. To achieve this, a new parallel corpus has been created by combining different available corpora online with preprocessing and cleaning. In addition, a monolingual Cantonese dataset has been created through web scraping to aid the synthetic parallel corpus generation. Following the data collection process, several approaches, including fine-tuning models, back-translation , and model switch , have been used. The translation quality of models has been evaluated with multiple quality metrics, including lexicon-based metrics (SacreBLEU and hLEPOR) and embedding-space metrics (COMET and BERTscore). Based on the automatic metrics, the best model is selected and compared against the 2 best commercial translators using a new human evaluation framework HOPES . The best model proposed in this investigation (NLLB-mBART) with model switch mechanisms has reached comparable and even better automatic evaluation scores against State-of-the-art commercial models (Bing and Baidu Translators), with a SacreBLEU score of 16.8 on our test set. Furthermore, an open-source web application has been developed to allow users to translate between Cantonese and English, with the different trained models available for effective comparisons between models from this investigation and users. CantonMT is available at https://github.com/kenrickkung/CantoneseTranslation