Open Software Development Research Articles

Distinction and alternative tech: Exploring the techno-critical disposition

How should we understand alternative social media and open-source technologies that seek to challenge the dominance of Big Tech? Are these ethical substitutes for monopolistic platforms and technological infrastructures, or “alternative” in the sense we might talk of alternative forms of culture? Here we offer new perspective on these questions by conceptualizing alternative tech through Bourdieu’s theories of cultural production and distinctive consumption. Building on the work of Holm, Coleman and others, we explore the “techno-critical disposition” through a case study of A Traversal Network of Feminist Servers (ATNOFS), arguing this is manifested primarily as “critical craftiness,” or hacker aesthetics in a critical register. Finally, we consider how ATNOFS represents a “distinctive” path to the wider adoption of alternative platforms, as well as how the techno-critical disposition may be reconfiguring legitimacy in the broader field of technology production.

How to unite local initiatives for a more sustainable global future

This article challenges the belief in high-tech solutions to solve socio-environmental crises, proposing a political vision beyond "green growth" and "ecomodernism.". It advocates for a commons-based technology framework, promoting collective resource management for sustainability. We thus introduce "cosmolocal" production, a configuration that strives to connect communities around shared resources and serve their needs while minimizing ecological impact. Despite acknowledged tensions, we contend that the cosmolocal framework could foster institutional and social change, aiming to address environmental degradation and wealth inequality. To support this contention on cosmolocal production's potential, we point to several successful examples from the open-source technology paradigm.

SHERPA-Cloud: An open-source online model to simulate air quality management policies in Europe

In this manuscript we present the SHERPA-Cloud online simplified air quality model, which is an evolution of the previous offline SHERPA model. SHERPA-cloud is a meta-model of a full Chemical Transport Model, and can be used to screen how policies would impact air quality. The main advantages of SHERPA-cloud (in comparison to the previous model version) are its speed (5 time faster than the offline version), its accessibility (permanent online availability) and its easier and more cost-effective maintenance (automatic updates can be shared more easily, and tool developments are facilitated). From the IT infrastructure and technologies points of view, the SHERPA-Cloud is based on an in-house cloud system called BDAP (Big Data Analytic Platform) which provides a ’JupyterLab’ service and tools to create dashboards using ’Voilà’ technology. SHERPA-Cloud is a useful easy-to-use tool in the hands of policy makers, citizens, NGOs and industries, that allows for interactive and participative policy developments. Indeed, it allows to quickly quantify changes in air quality resulting from the implementation of emission reduction strategies, in terms of both geographical or sectoral policies.

Knowledge sharing in manufacturing using LLM-powered tools: user study and model benchmarking.

Recent advances in natural language processing enable more intelligent ways to support knowledge sharing in factories. In manufacturing, operating production lines has become increasingly knowledge-intensive, putting strain on a factory's capacity to train and support new operators. This paper introduces a Large Language Model (LLM)-based system designed to retrieve information from the extensive knowledge contained in factory documentation and knowledge shared by expert operators. The system aims to efficiently answer queries from operators and facilitate the sharing of new knowledge. We conducted a user study at a factory to assess its potential impact and adoption, eliciting several perceived benefits, namely, enabling quicker information retrieval and more efficient resolution of issues. However, the study also highlighted a preference for learning from a human expert when such an option is available. Furthermore, we benchmarked several commercial and open-sourced LLMs for this system. The current state-of-the-art model, GPT-4, consistently outperformed its counterparts, with open-source models trailing closely, presenting an attractive option given their data privacy and customization benefits. In summary, this work offers preliminary insights and a system design for factories considering using LLM tools for knowledge management.

MemoryBank: Enhancing Large Language Models with Long-Term Memory

Large Language Models (LLMs) have drastically reshaped our interactions with artificial intelligence (AI) systems, showcasing impressive performance across an extensive array of tasks. Despite this, a notable hindrance remains—the deficiency of a long-term memory mechanism within these models. This shortfall becomes increasingly evident in situations demanding sustained interaction, such as personal companion systems, psychological counseling, and secretarial assistance. Recognizing the necessity for long-term memory, we propose MemoryBank, a novel memory mechanism tailored for LLMs. MemoryBank enables the models to summon relevant memories, continually evolve through continuous memory updates, comprehend, and adapt to a user's personality over time by synthesizing information from previous interactions. To mimic anthropomorphic behaviors and selectively preserve memory, MemoryBank incorporates a memory updating mechanism, inspired by the Ebbinghaus Forgetting Curve theory. This mechanism permits the AI to forget and reinforce memory based on time elapsed and the relative significance of the memory, thereby offering a more human-like memory mechanism and enriched user experience. MemoryBank is versatile in accommodating both closed-source models like ChatGPT and open-source models such as ChatGLM. To validate MemoryBank's effectiveness, we exemplify its application through the creation of an LLM-based chatbot named SiliconFriend in a long-term AI Companion scenario. Further tuned with psychological dialog data, SiliconFriend displays heightened empathy and discernment in its interactions. Experiment involves both qualitative analysis with real-world user dialogs and quantitative analysis with simulated dialogs. In the latter, ChatGPT acts as multiple users with diverse characteristics and generates long-term dialog contexts covering a wide array of topics. The results of our analysis reveal that SiliconFriend, equipped with MemoryBank, exhibits a strong capability for long-term companionship as it can provide emphatic response, recall relevant memories and understand user personality.

SeqGPT: An Out-of-the-Box Large Language Model for Open Domain Sequence Understanding

Large language models (LLMs) have shown impressive abilities for open-domain NLP tasks. However, LLMs are sometimes too footloose for natural language understanding (NLU) tasks which always have restricted output and input format. Their performances on NLU tasks are highly related to prompts or demonstrations and are shown to be poor at performing several representative NLU tasks, such as event extraction and entity typing. To this end, we present SeqGPT, a bilingual (i.e., English and Chinese) open-source autoregressive model specially enhanced for open-domain natural language understanding. We express all NLU tasks with two atomic tasks, which define fixed instructions to restrict the input and output format but still ``open'' for arbitrarily varied label sets. The model is first instruction-tuned with extremely fine-grained labeled data synthesized by ChatGPT and then further fine-tuned by 233 different atomic tasks from 152 datasets across various domains. The experimental results show that SeqGPT has decent classification and extraction ability, and is capable of performing language understanding tasks on unseen domains. We also conduct empirical studies on the scaling of data and model size as well as on the transfer across tasks. Our models are accessible at https://github.com/Alibaba-NLP/SeqGPT.

Aboveground hydrogen storage – Assessment of the potential market relevance in a Carbon-Neutral European energy system

Hydrogen storage is expected to play a crucial role in the comprehensive defossilization of energy systems. In this context, the focus is typically on underground hydrogen storage (e.g., in salt caverns). However, aboveground storage, which is independent of geological conditions and might offer other technical advantages, could provide systemic benefits and, thereby, gain shares in the hydrogen storage market. Against this background, this paper examines the market relevance of aboveground compared to underground hydrogen storage. Using the open-source energy system model and optimization framework of Europe, PyPSA-Eur, the influence of geological independence, storage cost relations, and technical storage characteristics (i.e., efficiencies) on mentioned market relevance of aboveground hydrogen storage are investigated. Further, the expectable market relevance based on current cost projections for the future is assessed. The studies show that in terms of hydrogen capacities, aboveground hydrogen storage plays a considerably smaller role compared to underground hydrogen storage. Even when assuming comparatively low aboveground storage cost, it will not exceed 1.7% (1.9TWhH2,LHV) of total hydrogen storage capacities in a cost-optimal European energy system. Regarding the amounts of annually stored hydrogen, aboveground storage could play a larger role, reaching a maximum share of 32.5% (168TWhH2,LHVa-1) of total stored hydrogen throughout Europe. However, these shares are only achievable for low cost storage in particularly suited energy system supply configurations. For higher aboveground storage costs or lower efficiencies, shares drop below 10% sharply. The analysis identifies some especially influential factors for achieving higher market relevance. Besides storage costs, the demand-orientation of a particular aboveground storage system (e.g., hydrogen storage at demand pressure levels) plays an essential role in market relevance. Further, overall efficiency can be a beneficial factor. Still, current projections of future techno-economic characteristics show that aboveground hydrogen storage is too expensive or too inefficient compared to underground storage. Therefore, to achieve notable market relevance, rather drastic cost reductions beyond current expectations would be needed for all assessed aboveground hydrogen storage technologies.

Development of Two-Step Method for Fast Chloride MSR Neutronics

This work presents neutronics models of a small and large fast-spectrum molten chloride-salt reactor. The models are similar to designs being pursued by industry, and they may serve as generic preconceptual and simplified neutronics models that provide information for decision making in licensing-related areas. The two models were created using Serpent, a Monte Carlo neutron transport code, and Moltres, a neutron diffusion core simulator tool. Specifically, this study focused on exploring the applicability of diffusion theory to fast molten salt reactor (MSR) models, the capabilities of an open-source, MSR-oriented simulation tool (Moltres), and optimal energy-group structures. The proposed two-step method involves group-constant generation with Serpent and a multigroup diffusion solution by Moltres. Three energy-group structures were applied. The accuracy of the solutions was determined through comparisons between the two-step and Monte Carlo flux and multiplication factor solutions. The findings indicated diffusion theory captures neutronics with minimal error for the large MSR and yielded best results with the 27-group structure. The 27-group structure yielded an average group flux error below 2% and keff agreement between diffusion and transport solutions within 30 pcm. The accuracy of the two-step method decreased for the very small (high-leakage) fast chloride MSR, but the neutronics were captured acceptably well with the 33-group structure. In addition to exploring the capabilities of Moltres, this work contributes to the sparse literature involving open-source models of fast-spectrum MSRs. Future work is noted as expanding the capabilities of the neutronics models to incorporate thermal hydraulics.

Abstract 3475: Prediction of nausea or vomiting, and fatigue or malaise in cancer care

Abstract Introduction: Natural language processing has recently achieved unprecedented performances for several medical tasks, but requires additional improvements in oncology. Moreover, very few projects have assessed the potential of language models to help prevent the most frequent serious or severe medical events in medical oncology. We aim at predicting nausea or vomiting (ICD10 code R11), and fatigue or malaise (ICD10 code R53) from patients' medical reports. Material and methods: The study included all the patients of Centre Léon Bérard between 2000 and 2023 that have not refused to share their data for this analysis. We have retrieved all the clinical notes and manual coding of hospitalization stays in ICD10, in French. We have pretrained a BERT language model with a masking strategy on this data and then fine-tuned it and compared it to several medical pretrained open source models (DrBert and K-memBERT). The labels were medical events leading to or associated with a hospitalization in the 90 days after every patients’ notes. For OncoBERT, we included sequential reports from the patients’ history, along with a time-encoding layer, and integrated it in a final transformer layer. Results: We analyzed 140,523 patients, representing 2,515,957 pseudo-anonymized text reports and 6.6M hospitalizations codes in total. The medical texts were consultations reports (56%), end-of-stay reports (17%) and hospitalization summaries (9%). The most frequent types of oncology treatments received by the patients at each time point were 18.8% for chemotherapy, 10.8% for targeted therapies and 1.3% for immunotherapies. The most frequent medical events were nausea or vomiting (20% of patients with 1 or more events), and fatigue or malaise (18% of patients with 1 or more events). In the final dataset, nausea and vomiting (R11) accounted for 16% of the labels while malaise and fatigue (R53) accounted for 24.7% of the labels. We performed random undersampling of reports without any event to balance the label dataset. The fine-tuning on R11 and R53 achieved the performances of 0.58 macro-aucpr (OncoBERT) and 0.50 macro-aucpr (best open-source model) on the validation set. Conclusion: The language models achieved high performance on the prediction of the most frequent serious medical events in our hospital dedicated to cancer care. We plan to validate the external performances of these models on collaborating hospitals and prospectively and improve the interpretations that we will present at the congress. Citation Format: Raphael Vienne, Quentin Filori, Vincent Susplugas, Hugo Crochet, Loic Verlingue. Prediction of nausea or vomiting, and fatigue or malaise in cancer care [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3475.

Abstract 4932: UltiAnalyzer.AI: An automatic and robust AI-Driven tool for the quantification of multiplex immunofluorescence whole slide images

Abstract Introduction: Whole slide multiplexed immunofluorescence (mIF) facilitates in-depth study of the tumor microenvironment with multi-channel images resulting in an enormous amount of protein expression and spatial data across millions of single cells. Therefore, it has become crucial to process this huge amount of data automatically and reliably. Here we introduce UltiAnalyzer.AI, a software tool that integrates a set of artificial intelligence (AI) models specifically designed for the analysis of whole slide mIF slides, that is fully automatic, highly accurate, highly efficient, and fully scalable. Methods: UltiAnalyzer.AI processes are specialized for 1) tissue segmentation, in which tissue regions are identified using a semantic segmentation AI model on low magnification versions of the DAPI channel, 2) tiling, where slide images are divided into 512x512 overlapping tiles for efficient image processing and lower memory requirements, 3) tumor region segmentation based on a semantic segmentation AI model, 4) nucleus segmentation using the open-source StarDist model, and 5) marker-positive cell detection, in which AI object detection models are employed to classify nuclei as positive or negative in each marker channel using the Faster R-CNN architecture. Here, we take advantage of the full 16-bit dynamic range of images produced using Ultivue InSituPlex® amplification technology, although our approach could be applied to other platforms. Intensity thresholding is no longer necessary, and identification of a positive cell mainly considers the morphology of the cell in the context of its cellular neighborhood. Results: Trained on over 60 mIF slides and validated with manual annotations, UltiAnalyzer.AI has demonstrated strong performance, achieving f1-scores between 0.8 to 0.91 for cell detection, and average correlations of 0.9 for cell counts across 15 markers. Processing time is around 3s per tile, enabling the processing of large slides in under an hour. It has been tested on over 1000 slides, and in a comparative study, it matched a VisioPharm-based quantification with 0.95 correlation in positive counts for 8 markers. Cloud deployment allows for processing thousands of slides simultaneously and in considerably less time. Conclusions: UltiAnalyzer.AI represents a significant advancement in the field of whole-slide mIF analysis, offering high accuracy and scalability using fully automated processes. By utilizing the full dynamic range of Ultivue’s mIF technology and avoiding traditional thresholding methods, the tool also increases robustness and reproducibility. The incorporation of cutting-edge AI technology and scalable cloud computing capabilities positions UltiAnalyzer.AI as a leading-edge solution for mIF analysis in digital pathology. Citation Format: Ruben Cardenes, Douglas Wood, Martin Schulze, Je Lee, Lorenz Rognoni. UltiAnalyzer.AI: An automatic and robust AI-Driven tool for the quantification of multiplex immunofluorescence whole slide images [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4932.

The potential of decentral heat pumps as flexibility option for decarbonised energy systems

Decarbonising the energy system requires high shares of variable renewable generation and sector coupling like power to heat. In addition to heat supply, heat pumps can be used in future energy systems to provide flexibility to the electricity system by using the thermal storage potential of the building stock and buffer tanks to shift electricity demand to hours of high renewable electricity production. Bridging the gap between two methodological approaches, we coupled a detailed building technology operation model and the open-source energy system model Balmorel to evaluate the flexibility potential that decentral heat pumps can provide to the electricity system. Austria in the year 2030 serves as an example of a 100% renewable-based electricity system (at an annual national balance). Results show that system benefits from heat pump flexibility are relatively limited in extent and concentrated on short-term flexibility. Flexible heat pumps reduce system cost, CO2 emissions, and photovoltaics and wind curtailment in all scenarios. The amount of electricity shifted in the assessed standard flexibility scenario is 194 GWhel and accounts for about 20% of the available flexible heat pump electricity demand. A comparison of different modelling approaches and a deterministic sensitivity analysis of key input parameters complement the modelling. The most important input parameters impacting heat pump flexibility are the flexible capacity (determined by installed capacity and share of control), shifting time limitations, and cost assumptions for the flexibility provided. Heat pump flexibility contributes more to increasing low residual loads (up to 22% in the assessed scenarios) than decreasing residual load peaks. Wind power integration benefits more from heat pump flexibility than photovoltaics because of the temporal correlation between heat demand and wind generation.

Comparative Analysis of Multimodal Large Language Model Performance on Clinical Vignette Questions

This study compares 2 large language models and their performance vs that of competing open-source models.

Emotion recognition based on microstate analysis from temporal and spatial patterns of electroencephalogram.

Recently, the microstate analysis method has been widely used to investigate the temporal and spatial dynamics of electroencephalogram (EEG) signals. However, most studies have focused on EEG at resting state, and few use microstate analysis to study emotional EEG. This paper aims to investigate the temporal and spatial patterns of EEG in emotional states, and the specific neurophysiological significance of microstates during the emotion cognitive process, and further explore the feasibility and effectiveness of applying the microstate analysis to emotion recognition. We proposed a KLGEV-criterion-based microstate analysis method, which can automatically and adaptively identify the optimal number of microstates in emotional EEG. The extracted temporal and spatial microstate features then served as novel feature sets to improve the performance of EEG emotion recognition. We evaluated the proposed method on two publicly available emotional EEG datasets: the SJTU Emotion EEG Dataset (SEED) and the Database for Emotion Analysis using Physiological Signals (DEAP). For the SEED dataset, 10 microstates were identified using the proposed method. These temporal and spatial features were fed into AutoGluon, an open-source automatic machine learning model, yielding an average three-class accuracy of 70.38% (±8.03%) in subject-dependent emotion recognition. For the DEAP dataset, the method identified 9 microstates. The average accuracy in the arousal dimension was 74.33% (±5.17%) and 75.49% (±5.70%) in the valence dimension, which were competitive performance compared to some previous machine-learning-based studies. Based on these results, we further discussed the neurophysiological relationship between specific microstates and emotions, which broaden our knowledge of the interpretability of EEG microstates. In particular, we found that arousal ratings were positively correlated with the activity of microstate C (anterior regions of default mode network) and negatively correlated with the activity of microstate D (dorsal attention network), while valence ratings were positively correlated with the activity of microstate B (visual network) and negatively correlated with the activity of microstate D (dorsal attention network). In summary, the findings in this paper indicate that the proposed KLGEV-criterion-based method can be employed to research emotional EEG signals effectively, and the microstate features are promising feature sets for EEG-based emotion recognition.

An Effective and Affordable Internet of Things (IoT) Scale System to Measure Crop Water Use

Scales are widely used in many agricultural applications, ranging from weighing crops at harvest to determine crop yields to regularly weighing animals to determine growth rate. In agricultural research applications, there is a long history of measuring crop water use (evapotranspiration [ET]) using a particular type of scale called weighing lysimeters. Typically, weighing lysimeters require very accurate data logging systems that tend to be expensive. Recent developments in open-source technologies, such as micro-controllers and Internet of Things (IoT) platforms, have created opportunities for developing effective and affordable ways to monitor crop water use and transmit the data to the Internet in near real-time. Therefore, this study aimed to create an affordable Internet of Things (IoT) scale system to measure crop ET. A scale system to monitor crop ET was developed using an Arduino-compatible microcontroller with cell phone communication, electronic load cells, an Inter-Integrated Circuit (I2C) multiplexer, and analog-to-digital converters (ADCs). The system was powered by a LiPo battery, charged by a small (6 W) solar panel. The IoT scale system was programmed to collect data from the load cells at regular time intervals and send the data to the ThingSpeak IoT platform. The system performed successfully during indoor and outdoor experiments conducted in 2023 at the Clemson University Edisto Research and Education Center, Blackville, SC. Calibrations relating the measured output of the scale load cells to changes in mass resulted in excellent linear relationships during the indoor (r2 = 1.0) and outdoor experiments (r2 = 0.9994). The results of the outdoor experiments showed that the IoT scale system could accurately measure changes in lysimeter mass during several months (Feb to Jun) without failure in data collection or transmission. The changes in lysimeter mass measured during that period reflected the same trend as concurrent soil moisture data measured at a nearby weather station. The changes in lysimeter mass measured with the IoT scale system during the outdoor experiment were accurate enough to derive daily and hourly crop ET and even detect what appeared to be dew formation during the morning hours. The IoT scale system can be built using open-source, off-the-shelf electronic components which can be purchased online and easily replaced or substituted. The system can also be developed at a fraction of the cost of data logging, communication, and visualization systems typically used for lysimeter and scale applications.

An open-source radar-based hail damage model for buildings and cars

Abstract. Severe hailstorms result in substantial damage to buildings and vehicles, necessitating the quantification of associated risks. Here, we present a novel open-source hail damage model for buildings and cars based on single-polarization radar data and 250 000 geolocated hail damage reports in Switzerland from 2002 to 2021. To this end, we conduct a detailed evaluation of different radar-based hail intensity measures at 1 km resolution and find that the maximum expected severe hail size (MESHS) outperforms the other measures, despite a considerable false-alarm ratio. Asset-specific hail damage impact functions for buildings and cars are calibrated based on MESHS and incorporated into the open-source risk modelling platform CLIMADA. The model successfully estimates the correct order of magnitude for the number of damaged building in 91 %, their total cost in 77 %, the number of damaged vehicles in 74 %, and their total cost in 60 % of over 100 considered large hail events. We found considerable uncertainties in hail damage estimates, which are largely attributable to limitations of radar-based hail detection. Therefore, we explore the usage of crowdsourced hail reports and find substantially improved spatial representation of severe hail for individual events. By highlighting the potential and limitations of radar-based hail size estimates, particularly MESHS, and the utilization of an open-source risk modelling platform, this study represents a significant step towards addressing the gap in risk quantification associated with severe hail events in Switzerland.

Improving the Capabilities of Large Language Model based Marketing Analytics Copilots with Semantic Search and Fine-Tuning

Artificial intelligence (AI) is widely deployed to solve problems related to marketing attribution and budget optimization. However, AI models can be quite complex, and it can be difficult to understand model workings and insights without extensive implementation teams. In principle, recently developed large language models (LLMs), like GPT-4, can be deployed to provide marketing insights, reducing the time and effort required to make critical decisions. In practice, there are substantial challenges that need to be overcome to reliably usesuch models. We focus on domain-specific question-answering, SQL generation needed for data retrieval, and tabular analysis and show how a combination of semantic search, prompt engineering, and fine-tuning can be applied to dramatically improve the ability of LLMs to execute these tasks accurately. We compare both proprietary models, like GPT-4, and open-source models, like Llama-2-70b, as well as various embedding methods. These models are tested on sample use cases specific to marketing mix modeling and attribution.

Sensitivity analysis of inverted model parameters from transient electromagnetic measurements affected by induced polarization effects

We investigate the application of the distance-based global sensitivity analysis (DGSA) to evaluate the sensitivity of electrical model parameters obtained from transient electromagnetic (TEM) data including induced polarization (IP) effects. We propose novel open-source forward modeling and inversion routines for single-loop TEM data including IP effects with the maximum phase angle model to model the frequency dependence of the complex resistivity. In a first step, we evaluate the accuracy of our forward modeling and inversion routines using numerical studies, where the actual variations of layer thicknesses and resistivities, as well as the frequency dependence of the complex resistivity is known. In a second step, we extend our investigation to field data and apply our approach to three distinct case studies in layered media: 1) a confined aquifer corresponding to conductive non-polarizable media, 2) a graphite deposit corresponding to highly conductive and polarizable anomalies in a resistive host rock and 3) an ice glacier corresponding to highly resistive polarizable media. Our DGSA results reveal that standard depth of investigation (DOI) approaches may overestimate the true sensitivity of the model obtained from the inversion. TEM data collected in conductive layered media without IP effects show reduced sensitivity above the predicted DOI. The case studies in polarizable media demonstrate that the maximum phase angle is more influential on the TEM model response than the relaxation time and dispersion coefficient. Our DGSA results for polarizable media reveal that TEM field data collected at the graphite deposit and at the ice glacier are sensitive to the geometry of the polarizable layer.

Electro-thermal performance evaluation of a prismatic battery pack for an electric vehicle

Abstract In recent years, electric vehicles (EVs) have grown in popularity as a viable way to reduce greenhouse gas emissions by replacing conventional vehicles. The need for EV batteries is steadily increasing. An essential and expensive part of electric transportation is the battery. The operating temperature of the lithium-ion (Li-ion) battery significantly impacts the performance of the EV battery pack. Battery packs undergo temperature fluctuations during the charging and discharging procedures due to internal heat generation, necessitating an examination of the temperature distribution of the battery pack. The geometrical spacing between cells is considered larger and identical and is kept open on two sides for free air circulation. A novel battery thermal management system (BTMS) design is required to effectively dissipate heat from the prismatic battery pack module. The electro-thermal behaviour of the prismatic Li-ion battery pack module was investigated based on the high charge/discharge rate. This study presents the development of a three-dimensional free open-source OpenFOAM computational fluid dynamics model for prismatic cell battery packs that simulates heat generation, air flow field, and temperature distribution across the width and depth of the battery pack module. The prismatic battery pack simulation results are compared with the experimental and simulation results of the cylindrical battery pack. It was also revealed that prismatic cells generate more heat on the backside, requiring battery packs to have increased cooling and space between individual cells to ensure sufficient air circulation for cooling and gas removal. The BTMS is improved by designing with increased space among the prismatic battery cells as compared with the conventional prismatic cell battery pack design.

Distributed Traffic Signal Optimization at V2X Intersections

This paper presents our research on a traffic signal control system (TSCS) at V2X intersections. The overall objective of the study is to create an implementable TSCS. The specific objective of this paper is to investigate a distributed system towards implementation. The objective function of minimizing queue delay is formulated as the integral of queue lengths. The discrete queueing estimation is mixed with macro and micro traffic flow models. The novel proposed architecture alleviates the communication network bandwidth constraint by processing BSMs and computing queue lengths at the local intersection. In addition, a two-stage distributed system is designed to optimize offsets, splits, and cycle length simultaneously and in real time. The paper advances TSCS theories by contributing a novel analytic formulation of delay functions and their first degree of derivatives for a two-stage optimization model. The open-source traffic simulation engine Enhanced Transportation Flow Open-Source Microscopic Model (ETFOMM version 1.2) was selected as a simulation environment to develop, debug, and evaluate the models and the system. The control delay of the major direction, minor direction, and the total network were collected to assess the system performance. Compared with the optimized TSCS timing plan by the Virginia Department of Transportation, the system generated a 21% control delay reduction in the major direction and a 7% control delay reduction in the minor direction at just a 10% penetration rate of connected vehicles. Finally, the proposed distributed and centralized systems present similar performances in the case study.

(109) Assessment of Artificial Intelligence and ChatGPT Quality in the Management of Erectile Dysfunction

Abstract Introduction Since November 2022, Artificial Intelligence (AI) chatbots have grown in popularity including urologic conditions. However, its accuracy and quality has not been evaluated systematically. In this study, we sought to assess the accuracy and quality of AI chatbots in the management of Erectile Dysfunction (ED). Objective We aim to evaluate the accuracy and quality of open-source language models in fielding common clinical questions pertaining to ED compared to board certified urologists. Methods Two AI open-source language models, ChatGPT and Google Bard, were fielded 15 standard questions related to ED some of which included causes, risk factors and treatment options of ED. Two board certified urologists were given the same questions on a standard survey. A third blinded board-certified urologist served as a grader using the AUA guidelines for ED on a Likert scale to assess the accuracy, robustness, and bias of each response. Urologist and AI responses were graded and aggregated using Likert scales. Results Overall AI responses were significantly more accurate (p<0.01), robust (p<0.01), and unbiased (p<0.01). Additionally, Google Bard had the highest scores all around followed by ChatGPT. The urologists’ responses were approximately 38% lower compared to the AI responses. Conclusions This study suggests that AI responses were superior compared to urologists in the areas of accuracy, robustness, and bias pertaining to the management of ED. Albeit, the chatbots have promising role in urologic conditions including ED, its widespread clinical use and adoption warrants further evaluation in the context of clinical decision making and enhancing patient care. Disclosure No.