Omar Costilla-Reyes, PhD

#AI is a tool. How will we use it for #education? On his visit to Indiana University School of Public Health-Bloomington, distinguished educator Marcio A. Oliveira, Ph D. described emerging uses (and misues) of AI in education.

20

Thrilled to announce that another paper titled as: "Optimized Neural Networks for Diabetes Classification Using Pima Indians Diabetes Database," is now published on IEEE Xplore! You can read it here:

22

3 Comments

Calling all aspiring #AI researchers and #NLP scientists! 🗣 🤖 Don't miss this new book 📖 , Natural Language Processing in Biomedicine, co-edited by Hua Xu, PhD, Yale Biomedical Informatics & Data Science vice chair and Yale School of Medicine assistant dean for biomedical informatics. The book includes topics ranging from biomedical #ML to #LLM #genai and more. In our summer newsletter feature, read more about this new essential reading for students and experts alike: https://lnkd.in/eJX3yXHc

134

5 Comments

Welcome to read and share the newly published paper "Inertial Sensors-Based Assessment of Human Breathing Pattern: A Systematic Literature Review". This excellent paper is written by Rodrigo Martins, Fátima Rodrigues, Susana Costa and Nélson Costa. Read via: https://lnkd.in/gREA6HPV MDPI Universidade do Minho #machinelearning #BreathingPattern

3

Predictive modeling of biomedical temporal data in healthcare applications: review and future directions

5

AI and the 2024 Nobel Prizes in Physics and Chemistry Prof. Dr. Maurício Veloso Brant Pinheiro, Departamento de Física UFMG The 2024 Nobel Prize recognizes transformative advances in Artificial Intelligence (AI) in the fields of Physics and Chemistry, with revolutionary implications for science and medicine. Figures like Geoffrey Hinton and John Hopfield, pioneers in machine learning, and Demis Hassabis, John Jumper, and David Baker, who used AI to uncover protein structures, demonstrate how these innovations are shaping the future of scientific research and the development of medical treatments. In addition to pushing the boundaries of science, these achievements also spark fundamental debates on AI ethics and regulation, heralding a promising future where technology and science intertwine in innovative and responsible ways. https://lnkd.in/dwszAbYp #AlphaFold #DeepMind #DemisHassabis #GeoffreyHinton #AI #ArtificialIntelligence #JohnHopfield #JohnJumper #MachineLearning #NeuralNetwork #Nobel2024 #NobelPrize #Google

1

Fort Collins (and nearby places) friends and colleagues. We would appreciate your help. I've included the message from my student below. IRB #5732 protocol approved. Dear all, My name is Lena Plabst, and I work with Dr. Francisco R. Ortega (PI). We are researchers from Colorado State University's Computer Science Department. We are recruiting human participants for an eye-tracking experiment called Visual Interfaces with Text with Gaze Tracking, where you will be asked to select text labels on several images. These text labels will be presented with different colors, outlines, etc. During the experiment, your gaze will be measured using an eye tracker to determine your behavior while identifying these labels. The eye tracker will first have to be calibrated, for which you will be asked to follow a dot with your eyes. The experiment takes 90 minutes. There are no direct benefits to you as a participant, but you will be able to gain some understanding of eye-tracking research. You may experience eyestrain like the eyestrain experienced during normal computer usage. You may experience slight discomfort from having to hold the same position for an extended period of time. The study aims to understand how to make preliminary predictions about the features contributing to saliency in images involving synthetically inserted text and how the presence of synthetically inserted text overlays will change how the participant engages with the image. As compensation for your participation in the study, you will receive a $20 amazon Gift Cards. Please note that you can decline the compensation. Please sign up for this study if the following applied to you: - Over the age of 18 - Normal or corrected to normal vision - English as a first language Please register for an experiment slot. Use this link: https://lnkd.in/gZSkB2fA The experiment will take place at Colorado State University Fort Collins, CO, Computer Science Building 377. You will inform us of your wish to participate, and we will contact you regarding the time. If you have any questions, feel free to email me at [email protected] and/or [email protected]. If you have any questions about your rights as a volunteer in this research, contact the CSU IRB at [email protected]; 970-491-1553. Thanks,

17

Study of influence of Large Language Models (LLMs) on diagnostic reasoning for a set of clinical vignettes. 50 participating clinicians randomized into two groups: (1) a group using only conventional informational resources and (2) a group with access to LLMs + conventional resources. Findings: Providing an LLM to physicians did not significantly improve clinical reasoning vs. those using only conventional resources. And the LLM alone demonstrated higher performance than the two physician groups. The results may show that (1) physicians would benefit from training to better harness LLMs for diagnosis, and (2) much can be done with innovation with clinician-AI collaboration and complementarity. H/T Eric Horvitz

16

This is a nice paper from a young PhD student in Morocco, Mohammed K.. The focus is on the use of #virtualreality in #healthcare #learning and #education. The specific use case is #anatomy and a test of the Immersive Virtual Anatomy Laboratory (IVAL). IVAL combines virtual reality and serious games to enhance anatomy education. The study compared IVAL with traditional learning methods regarding educational effectiveness and user acceptance. 120 undergraduate health-science students were participants and were randomly assigned to two groups: Experimental Group: Used IVAL. Control Group: Followed traditional learning methods. Data Collection: Quantitative: Pretest and posttest vocabulary scores, task completion times. Qualitative: User experience questionnaire. Framework: Utilizes Technology Acceptance Model (TAM) focusing on Perceived Usefulness and Perceived Ease of Use. Results: Vocabulary Scores: Experimental group saw a 55.95% increase. Task Completion Times: Experimental group experienced an 18.75% reduction. Qualitative feedback indicated higher Perceived Usefulness, improved Perceived Ease of Use, positive attitudes towards IVAL, and strong Behavioral Intention to continue using IVAL. Conclusion: IVAL shows promise in enhancing anatomy education effectiveness and enjoyment; indicates the potential for immersive learning environments to improve learning outcomes and user acceptance. Really nice work and an important application of #immersivetechnology!

72

8 Comments

We are delighted to share you the paper recently published in Medinformatics! The paper is titled "Exploring Disrupted Gene Networks in Human 22q11.2 Microdeletion". It is co-authored by Camila Cristina de Oliveira Alves, Ivan Rodrigo Wolf, and Guilherme Targino Valente from Sao Paulo State University- The Universidade Estadual Paulista (UNESP) and Bruno Faulin Gamba, and Lucilene Arilho Ribeiro Bicudo from Universidade Federal de Goiás. Abstract: Several deletions are observed at the 22q11 locus and are responsible for 22q11.2 deletion syndrome (22q11DS), also known as DiGeorge syndrome, conotruncal anomaly face syndrome, or velocardiofacial syndrome. These microdeletions on human chromosome 22 range from 0.7 to 3 Mb. Many genes are affected by 22q11.2 deletion. However, despite the well-established clinical signs for the diagnosis of 22q11.2 deletion syndrome, the interactome background of 22q11.2 deletion syndrome is unknown. Here, we analyzed protein–protein interaction networks (PPIs) to assess the influences of 3 Mb 22q11.2 deletion on this network. We compared the general human PPI network against a network without 48 genes of the 3 Mb 22q11.2 locus in a homozygous condition: we compared topological metrics, enrichment of gene ontology terms, community assignments, and edge rewiring. The PPI networks revealed that this deletion affected the relevance of hundreds of nondeleted genes. Additionally, this 22q11.2 deletion induces intense rewiring of subnetworks, promoting an accumulation of proteins associated with DiGeorge clinical signs (CTCF, YY1, TFAP2A, PPARG, PAX6, RAX, and E2F3) in a single community (community 1). Therefore, we identified new genes that may be associated with the 22q11.2 deletion syndrome. Altogether, the systemic approaches used here yielded new insights into the 22q11.2 deletion syndrome. To read and share the full paper via the link: https://lnkd.in/gFMZcefM #bioinformatics #medicalresearch #systemsbiology 

10

In computational biology publishing a paper is only part of the journey - Alper Uzun, MS, PhD https://lnkd.in/dDGJB-Wf Jennifer Li Andy Yang Benedito A. Carneiro, MD, MS Ece Uzun, MS, Ph.D.,FAMIA Lauren Massingham,MD Brown University #Cancer #Bioinformatics #Genomics #ComputationalBiology #Medicine #CancerResearch #VariantAnalysis #VGC #ResearchTools #Health #Oncology #OncoDaily

16

🚀 Attention all software engineers! ACM has launched the first issue of Proceedings of the ACM on Software Engineering (PACMSE), a Gold Open Access journal dedicated to advancing software engineering research. PACMSE covers a wide range of topics, from design to deployment, aiming to improve software quality. Each issue of the journal comprises a curated selection of papers, which will be announced through publicized calls for papers. For more detail on the submission process, refer to intro in the inaugural issue: https://bit.ly/3YwVBti  📚 #SoftwareEngineering #OpenAccess #ACM

59

1 Comment

Interested in interpretability of AI? I suggest this overview paper where we present the many possibilities researchers have to approach it. Asking for interpretability is critical in health-related problems but I think would benefit all areas as well. https://lnkd.in/ePP8NPmX

20

3 Comments

I am excited to share that our paper, "Silver Lining in the Fake News Cloud: Can Large Language Models Help Detect Misinformation?" has been accepted for publication in IEEE Transactions on Artificial Intelligence (IEEE TAI)! Abstract: In the age of advanced generative AI, distinguishing truth from deception is a growing societal challenge. Our research explores how Large Language Models (LLMs) can aid in detecting misinformation. We use a comprehensive approach, evaluating multiple LLMs with few and zero-shot prompts and incorporating sentiment and emotion annotations to understand their impact on misinformation detection. Additionally, we deliberately distort news using ChatGPT to examine various parameters like abstractness and concreteness, enhancing our framework for identifying authentic, human-generated, and LLM-induced misinformation. Impact Statement: As LLMs evolve, they play a dual role as both defenders and perpetrators of misinformation. This research delves into the conflicting nature of LLMs, evaluating their effectiveness across diverse datasets and prompting techniques. By corrupting authentic news and human-generated fake news with ChatGPT, we quantify parameters like concreteness and abstractness to improve misinformation detection tools. Our findings aim to enhance current frameworks and contribute to the responsible development of AI, fostering a more reliable and resilient information ecosystem in the face of misinformation. Key Contributions: LLM Evaluation: Assessed multiple LLMs' effectiveness in detecting rumor content across six datasets. Prompting Techniques: Investigated few-shot and zero-shot prompting and examined sensitivity to hyperparameters. Sentiment & Emotion: Explored the impact of sentiment and emotion on LLMs' rumor detection capabilities. Dual Corruption Approach: Intentionally distorting both authentic news and human-generated misinformation using ChatGPT. We analyzed parameters such as concreteness and named entity density to gain deeper insights into the impact of these distortions. Explore our research and access the code on GitHub: https://lnkd.in/gbY-9QcT A huge thank you to my co-authors Raghvendra Kumar, Bhargav Goddu, Adam Jatowt, and all those who supported this journey. Here's to advancing the fight against misinformation with cutting-edge AI technologies! #Research #IEEE #ArtificialIntelligence #MisinformationDetection #LargeLanguageModels #AI #FakeNews #IEEE_TAI

250

11 Comments

Great to see TRIPOD+AI (& CONSORT-AI/SPIRIT-AI) mentioned in the MRC "Interim guidance for biomedical AI and software developers" to ensure tools are fit for purpose & the benefits available to all in society (Thanks Brendan Delaney)👇 https://lnkd.in/esVSFRZE TRIPOD+AI guidance is available here - https://lnkd.in/etCkc_Gp (NB don't forget the supplementary table 1, as it contains additional pointers on what needs reporting - whilst we finish up a more detailed guidance with examples in our TRIPOD+AI Explanation & Elaboration paper - coming in 2025) CONSORT-AI is available here - https://lnkd.in/eHTy-_hV SPIRIT-AI is available here - https://lnkd.in/eztvc_En PROBAST+AI should appear early next year.

49

1 Comment

This is tremendously exciting! The largest #OpenSource initiative in #ComputationalPathology by far. It includes the #Virchow #FoundationModel, trained on billions of images from 1.5 million slides, available under the Apache 2.0 license for everyone to use. On top of that, the #PRISM model, which provides whole-slide embeddings for pan-cancer detection, subtyping, and IHC biomarker prediction. Download them and try them out! It takes an entire field to revolutionize cancer research and make a tangible clinical impact on patient care.

146

3 Comments

Our paper on abstracting PubMed knowledge graphs for improving link prediction is available now. Usually link prediction involves point-to-point links between two entities, while there is rich community structure in PubMed. By first grouping entities of the same types (communities), we can overcome sparsity related issues when predicting links between entities. Joint work with the amazing folks at Deloitte. Yang Zhao, Shunsuke Hirose, Yingzi Jin Tomotake Kozu https://lnkd.in/eG8sX5gG

38

We have the following two papers accepted in EMNLP 2024 Findings. 1) HealthAlignSumm: Utilizing Alignment for Multimodal Summarization of Code-Mixed Healthcare Dialogues by Akash Ghosh, Arkadeep Acharya, Sriparna Saha, Gaurav Pandey, Dinesh Raghu, Setu Sinha Contributions: a) This study pioneers the Multimodal Dialog Summarization Task within a Hindi-English 120 codemixed environment, representing a novel initiative in healthcare research. By embracing linguistic diversity, our work sets a precedent for enhanced personalization and seamless communication in healthcare settings. b) We curated a novel dataset, namely Multimodal Codemixed Dialogue Summarization in Healthcare (MCDH), to advance research in this direction. We utilized a novel automated approach, utilizing LLM technology, to simplify the creation of the Hinglish dataset. Furthermore, we conducted a comprehensive analysis of the dataset to ensure its linguistic quality and suitability for research purposes. c) We propose HealthAlignSumm, an innovative model that integrates visual cues into the encoder-decoder transformer architecture. It introduces a contextual image fusion layer in the encoder and an image cross-attention layer in the decoder, pioneering image fusion in BART’s decoder. Using Direct Preference Optimization with an LLM-generated preference dataset, our model aligns effectively. This is the first work to show that alignment with synthetically generated preference datasets can refine outputs for multimodal summary generation. 2) Unveiling Hallucination in Text, Image, Video, and Audio Foundation Models: A Comprehensive Review by Pranab Sahoo, Prabhash Meharia, Akash Ghosh, Sriparna Saha, Vinija Jain, Aman Chadha Contributions: (a) This survey provides a comprehensive overview of hallucination detection and mitigation techniques designed specifically for multimodal foundation models, extending beyond the typical focus solely on language models. (b) Establish a precise definition and structured taxonomy of hallucination in the context of large-scale foundation models. (c) We have presented the various detection and mitigation strategies that have been proposed to address the hallucination problem in a multimodal setting. (d) Highlight the open challenges and future research directions in this critical area. Heartiest congratulations to all authors!

257

9 Comments