News & Updates
October 17, 2024
Challenges for AI: Detecting Manipulative Language in Conversations
Yuxin Wang, a Dartmouth PhD student, is exploring the capacity of large language models, such as GPT-4 and Meta's Llama-2, to identify manipulative language in digital communications. Presented at the Association for Computational Linguistics, her research with the MentalManip dataset reveals that these models often misinterpret manipulation as general toxicity, underscoring the difficulty AI faces in recognizing subtle forms of verbal abuse. This study highlights the need for specialized AI tools to accurately detect and manage nuanced interpersonal manipulations.
https://web.cs.dartmouth.edu/news/2024/10/can-large-language-models-identify-manipulative-language
September 22, 2024
AI Advancement in Endometrial Cancer: The EndoNet Model
Researchers and investigators at CPHAI have introduced EndoNet, an innovative AI model that enables endometrial cancer grading without the need for labor-intensive manual annotation by pathologists. By leveraging convolutional neural networks and vision transformers, EndoNet meticulously analyzes histological features from whole-slide images with high sensitivity and specificity. Validated using data from The Cancer Genome Atlas, this novel tool promises to significantly improve diagnostic accuracy in gynecological pathology, thereby assisting pathologists in their diagnostic process.
September 12, 2024
Introducing MENTALMANIP: A Dataset for Understanding Conversational Manipulation
Dartmouth researchers have created MENTALMANIP, a vital resource aimed at studying mental manipulation in conversations in order to improve mental health interventions. This dataset, with 4,000 annotated dialogues, illuminates the tactics of manipulation and the vulnerabilities of victims. Despite using sophisticated models, accurately detecting manipulative language remains challenging even for advanced AI systems.
September 2, 2024
Introducing MaskHIT: A Foundational Model for Pathology Image Analysis
In the latest advancements at CPHAI, researchers have developed the foundational model, named MaskHIT, transforming digital pathology with its innovative approach to analyzing whole-slide images (WSIs). Trained on over 7,000 whole-slide images, MaskHIT excels in identifying critical histological patterns, outperforming traditional methods in survival predictions and cancer classification. By learning directly from image features without extensive labeled datasets, MaskHIT aligns with clinician annotations, making it a valuable resource for accurately identifying histological structures and informing patient treatments.
July 24, 2024
Director’s Interview with Connections Magazine “The Future of Healthcare: The Impact of AI“
In a recent discussion with Connections Magazine, the director of CPHAI shared insights on how AI could significantly transform healthcare. He explained that AI has the potential to make healthcare more tailored and proactive, helping to diagnose diseases earlier and manage treatment more effectively. Emphasizing the importance of using AI responsibly, he highlighted the center's commitment to ensuring that AI benefits everyone equitably and ethically.
https://www.dartmouth-hitchcock.org/stories/article/future-healthcare-impact-ai
August 23, 2024
Advancing AI Transparency with a National Registry
In a recent interview with STAT News, CPHAI's Director discussed the complexities of implementing a national AI registry. He supports standardized tracking and monitoring of AI technologies in the medical domain due to their significant health and safety implications. However, he emphasizes that any systematic approach must account for all stakeholders' diverse needs and resources, including small and large hospitals and AI vendors. He highlights the necessity for a comprehensive benchmarking mechanism to address the diverse requirements, ensuring effective deployment of AI models across varied healthcare environments
https://www.statnews.com/2024/08/23/artificial-intelligence-health-care-national-registry/
May 26, 2024
CPHAI Symposium Recap
This spring, the Dartmouth Center for Precision Health and Artificial Intelligence held its inaugural symposium at the Hanover Inn, attracting a diverse audience to explore AI in healthcare. President Sian Beilock opened the event, emphasizing Dartmouth's unique role in fostering interdisciplinary collaboration to advance AI applications in healthcare. She highlighted ongoing research and the potential of AI to improve well-being and mental health. The Director of the Center, Saeed Hassanpour, along with keynote speakers Dr. Curtis Langlotz from Stanford University and Dr. Faisal Mahmood from Harvard University, discussed the impact of AI on diagnostics and treatment, stressing the importance of responsible innovation.
May 16, 2024
Collaboration Portal Launch at CPHAI’s Website
We are excited to announce a new section on our website dedicated to supporting collaborations on AI and precision health. Here, students, faculty, clinicians, and research staff interested in integrating AI into healthcare solutions can reach out to Center members. CPHAI will consult and brainstorm various ideas, connect them with the required clinical or computational expertise, and facilitate the implementation and execution of novel ideas to achieve impactful and transformative outcomes in healthcare.
May 12, 2024
Advancing NLP: Transparent AI with Proto-LM
Proto-LM is a novel tool developed by collaborators at CPHAI that aims to improve the understandability of large language models (LLMs) without compromising performance. This tool enhances how these models are fine-tuned, providing clear explanations of their decisions. It is a step toward making AI more transparent and trustworthy, ensuring that the models we rely on can be both effective and accountable.
February 21, 2024
CPHAI's Inaugural Symposium on AI in Medicine: Bridging Innovation & Practice
CPHAI will host its inaugural symposium on Precision Health & AI at Dartmouth on April 3rd in the Hanover Inn Grand Ballroom. This event will feature Dartmouth's AI experts and clinicians discussing their pioneering research and advances. The CPHAI Symposium will be inaugurated with remarks from President Sian Beilock, featuring Dr. Curtis Langlotz of Stanford and Dr. Faisal Mahmood from Harvard Medical School, known for their novel research in medical imaging and pathology
September 26, 2023
CPHAI's Director to Speak at the Inaugural Dartmouth AI Conference
Our Center Director will present CPHAI at the inaugural Dartmouth AI Conference on September 29, 2023. This event, hosted by the Tuck School of Business, celebrates Dartmouth's pioneering role in AI. Dr. Hassanpour will underscore CPHAI's commitment to the future of precision health and AI at this meeting.
September 20, 2023
CPHAI's Director Joins Expert Panel at Post-COVID AI Symposium
Our Center Director will be participating in the "AI Expert Panel: AI futures" at the Post-COVID AI Symposium in Boston, MA. This symposium, scheduled for September 20th, 2023, focuses on the transformative role of AI in healthcare following the global challenges of the COVID-19 pandemic. Joining a panel of experts, Dr. Saeed Hassanpour will discuss the anticipated breakthroughs in AI and their potential impact on medicine.
November 3, 2023
New Faculty Position Announced in Precision Health
The Geisel School of Medicine at Dartmouth is excited to announce a new tenure-track faculty position in the Department of Biomedical Data Science and the Center for Precision Health & Artificial Intelligence. This role focuses on AI and machine learning in precision health, engaging in a highly collaborative and interdisciplinary environment with access to advanced research facilities. Candidates with a strong background and interest in AI and its applications in medicine are encouraged to apply.
April 4, 2024
CPHAI's Symposium Explores AI Revolution in Health Care
The one-day AI in Medicine: Bridging Innovation & Practice event, held at the Hanover Inn, highlighted AI-driven innovations that are already changing the landscape of health care and sparked thoughtful discussions on recognizing potential pitfalls that the technology presents.
https://dartmouth.edu/news/symposium-explores-ai-revolution-health-care
March 11, 2024
Navigating CRC Diagnosis: The Impact of Deep Learning on Early Detection
At Dartmouth, a recent work introduces an AI-powered approach to colorectal cancer (CRC) diagnostics using deep learning to analyze histopathologic slides. This method improves the accuracy in distinguishing between benign, low-grade, and high-grade dysplasia and adenocarcinoma for CRC patients. This research aims to offer quicker, more precise CRC detection, facilitating timely and effective patient treatment options
March 23, 2024
AI-Driven Insights: A Novel Biomarker for Breast Cancer Care
At Dartmouth, a recent work introduces an AI-powered approach to colorectal cancer (CRC) diagnostics using deep learning to analyze histopathologic slides. This method improves the accuracy in distinguishing between benign, low-grade, and high-grade dysplasia and adenocarcinoma for CRC patients. This research aims to offer quicker, more precise CRC detection, facilitating timely and effective patient treatment options
April 1, 2024
Dartmouth event to explore the promise – and pitfalls – of AI in medicine
The one-day AI in Medicine: Bridging Innovation & Practice event, held at the Hanover Inn, highlighted AI-driven innovations that are already changing the landscape of health care and sparked thoughtful discussions on recognizing potential pitfalls that the technology presents.
https://www.nhpr.org/health/ai-artificial-intelligence-medicine-healthcare-dartmouth
April 3, 2024
Dartmouth Symposium highlights AI’s innovations in medicine
A first-of-its-kind symposium Wednesday at Dartmouth College explored both the potential benefits and pitfalls of artificial intelligence in health care. Dr. Curt Langlotz, a professor at Stanford, is a leader in the AI field, investigating how machine learning is getting better at detecting diseases, specifically when it comes to reading X-rays and other scans.
May 6, 2024
Integrating AI in Medical Education: A New Approach at Dartmouth
At Dartmouth's Geisel School of Medicine, the CPHAI assisted in developing a pilot AI elective within the Digital Health Scholars program. This course integrates AI and digital health concepts into preclinical studies, enhancing medical students' ability to apply these technologies effectively. The elective aims to increase student confidence in AI, preparing them to utilize these tools in their future medical practices to improve diagnostics and patient care in a rapidly evolving healthcare environment.