Citations

In postoperative care, patient education and follow-up are pivotal for enhancing the quality of care and satisfaction. Artificial intelligence virtual assistants (AIVA) and large language models (LLMs) like Google BARD and ChatGPT-4 offer avenues for addressing patient queries using natural language processing (NLP) techniques. However, the accuracy and appropriateness of the information vary across these platforms, necessitating a comparative study to evaluate their efficacy in this domain. We conducted a study comparing AIVA (using Google Dialogflow) with ChatGPT-4 and Google BARD, assessing the accuracy, knowledge gap, and response appropriateness. AIVA demonstrated superior performance, with significantly higher accuracy (mean: 0.9) and lower knowledge gap (mean: 0.1) compared to BARD and ChatGPT-4. Additionally, AIVA’s responses received higher Likert scores for appropriateness. Our findings suggest that specialized AI tools like AIVA are more effective in delivering precise and contextually relevant information for postoperative care compared to general-purpose LLMs. While ChatGPT-4 shows promise, its performance varies, particularly in verbal interactions. This underscores the importance of tailored AI solutions in healthcare, where accuracy and clarity are paramount. Our study highlights the necessity for further research and the development of customized AI solutions to address specific medical contexts and improve patient outcomes.

This study aims to explore how artificial intelligence can help ease the burden on caregivers, filling a gap in current research and healthcare practices due to the growing challenge of an aging population and increased reliance on informal caregivers. We conducted a search with Google Scholar, PubMed, Scopus, IEEE Xplore, and Web of Science, focusing on AI and caregiving. Our inclusion criteria were studies where AI supports informal caregivers, excluding those solely for data collection. Adhering to PRISMA 2020 guidelines, we eliminated duplicates and screened for relevance. From 947 initially identified articles, 10 met our criteria, focusing on AI’s role in aiding informal caregivers. These studies, conducted between 2012 and 2023, were globally distributed, with 80% employing machine learning. Validation methods varied, with Hold-Out being the most frequent. Metrics across studies revealed accuracies ranging from 71.60% to 99.33%. Specific methods, like SCUT in conjunction with NNs and LibSVM, showcased accuracy between 93.42% and 95.36% as well as F-measures spanning 93.30% to 95.41%. AUC values indicated model performance variability, ranging from 0.50 to 0.85 in select models. Our review highlights AI’s role in aiding informal caregivers, showing promising results despite different approaches. AI tools provide smart, adaptive support, improving caregivers’ effectiveness and well-being.

Background: OpenAI’s ChatGPT (San Francisco, CA, USA) and Google’s Gemini (Mountain View, CA, USA) are two large language models that show promise in improving and expediting medical decision making in hand surgery. Evaluating the applications of these models within the field of hand surgery is warranted. This study aims to evaluate ChatGPT-4 and Gemini in classifying hand injuries and recommending treatment. Methods: Gemini and ChatGPT were given 68 fictionalized clinical vignettes of hand injuries twice. The models were asked to use a specific classification system and recommend surgical or nonsurgical treatment. Classifications were scored based on correctness. Results were analyzed using descriptive statistics, a paired two-tailed t-test, and sensitivity testing. Results: Gemini, correctly classifying 70.6% hand injuries, demonstrated superior classification ability over ChatGPT (mean score 1.46 vs. 0.87, p-value < 0.001). For management, ChatGPT demonstrated higher sensitivity in recommending surgical intervention compared to Gemini (98.0% vs. 88.8%), but lower specificity (68.4% vs. 94.7%). When compared to ChatGPT, Gemini demonstrated greater response replicability. Conclusions: Large language models like ChatGPT and Gemini show promise in assisting medical decision making, particularly in hand surgery, with Gemini generally outperforming ChatGPT. These findings emphasize the importance of considering the strengths and limitations of different models when integrating them into clinical practice.

Abstract Approximately 5% of Alzheimer’s disease patients develop symptoms before age 65 (early-onset Alzheimer’s disease), with either sporadic (sporadic early-onset Alzheimer’s disease) or dominantly inherited (dominantly inherited Alzheimer’s disease) presentations. Both sporadic early-onset Alzheimer’s disease and dominantly inherited Alzheimer’s disease are characterized by brain amyloid-β accumulation, tau tangles, hypometabolism and neurodegeneration, but differences in topography and magnitude of these pathological changes are not fully elucidated. In this study, we directly compared patterns of amyloid-β plaque deposition and glucose hypometabolism in sporadic early-onset Alzheimer’s disease and dominantly inherited Alzheimer’s disease individuals. Our analysis included 134 symptomatic sporadic early-onset Alzheimer’s disease amyloid-Positron Emission Tomography (PET)-positive cases from the University of California, San Francisco, Alzheimer’s Disease Research Center (mean ± SD age 59.7 ± 5.6 years), 89 symptomatic dominantly inherited Alzheimer’s disease cases (age 45.8 ± 9.3 years) and 102 cognitively unimpaired non-mutation carriers from the Dominantly Inherited Alzheimer Network study (age 44.9 ± 9.2). Each group underwent clinical and cognitive examinations, 11C-labelled Pittsburgh Compound B-PET and structural MRI. 18F-Fluorodeoxyglucose-PET was also available for most participants. Positron Emission Tomography scans from both studies were uniformly processed to obtain a standardized uptake value ratio (PIB50–70 cerebellar grey reference and FDG30–60 pons reference) images. Statistical analyses included pairwise global and voxelwise group comparisons and group-independent component analyses. Analyses were performed also adjusting for covariates including age, sex, Mini-Mental State Examination, apolipoprotein ε4 status and average composite cortical of standardized uptake value ratio. Compared with dominantly inherited Alzheimer’s disease, sporadic early-onset Alzheimer’s disease participants were older at age of onset (mean ± SD, 54.8 ± 8.2 versus 41.9 ± 8.2, Cohen’s d = 1.91), with more years of education (16.4 ± 2.8 versus 13.5 ± 3.2, d = 1) and more likely to be apolipoprotein ε4 carriers (54.6% ε4 versus 28.1%, Cramer’s V = 0.26), but similar Mini-Mental State Examination (20.6 ± 6.1 versus 21.2 ± 7.4, d = 0.08). Sporadic early-onset Alzheimer’s disease had higher global cortical Pittsburgh Compound B-PET binding (mean ± SD standardized uptake value ratio, 1.92 ± 0.29 versus 1.58 ± 0.44, d = 0.96) and greater global cortical 18F-fluorodeoxyglucose-PET hypometabolism (mean ± SD standardized uptake value ratio, 1.32 ± 0.1 versus 1.39 ± 0.19, d = 0.48) compared with dominantly inherited Alzheimer’s disease. Fully adjusted comparisons demonstrated relatively higher Pittsburgh Compound B-PET standardized uptake value ratio in the medial occipital, thalami, basal ganglia and medial/dorsal frontal regions in dominantly inherited Alzheimer’s disease versus sporadic early-onset Alzheimer’s disease. Sporadic early-onset Alzheimer’s disease showed relatively greater 18F-fluorodeoxyglucose-PET hypometabolism in Alzheimer’s disease signature temporoparietal regions and caudate nuclei, whereas dominantly inherited Alzheimer’s disease showed relatively greater hypometabolism in frontal white matter and pericentral regions. Independent component analyses largely replicated these findings by highlighting common and unique Pittsburgh Compound B-PET and 18F-fluorodeoxyglucose-PET binding patterns. In summary, our findings suggest both common and distinct patterns of amyloid and glucose hypometabolism in sporadic and dominantly inherited early-onset Alzheimer’s disease.

With abundant information and interconnectedness among people, identifying knowledgeable individuals in specific domains has become crucial for organizations. Artificial intelligence (AI) algorithms have been employed to evaluate the knowledge and locate experts in specific areas, alleviating the manual burden of expert profiling and identification. However, there is a limited body of research exploring the application of AI algorithms for expert finding in the medical and biomedical fields. This study aims to conduct a scoping review of existing literature on utilizing AI algorithms for expert identification in medical domains. We systematically searched five platforms using a customized search string, and 21 studies were identified through other sources. The search spanned studies up to 2023, and study eligibility and selection adhered to the PRISMA 2020 statement. A total of 571 studies were assessed from the search. Out of these, we included six studies conducted between 2014 and 2020 that met our review criteria. Four studies used a machine learning algorithm as their model, while two utilized natural language processing. One study combined both approaches. All six studies demonstrated significant success in expert retrieval compared to baseline algorithms, as measured by various scoring metrics. AI enhances expert finding accuracy and effectiveness. However, more work is needed in intelligent medical expert retrieval.

Abstract Background Two years of adjuvant abemaciclib + endocrine therapy (ET) resulted in significant and clinically meaningful improvement in invasive disease-free survival (IDFS) and distant relapse-free survival (DRFS) in patients (pts) with HR+, HER2-, node-positive, high-risk early breast cancer (EBC) in the monarchE trial (NCT03155997). The benefit of abemaciclib was sustained beyond completion of treatment and deepened in magnitude at 4 years for IDFS and DRFS. This pilot study investigated the technical feasibility of ctDNA detection beginning prior to study treatment, as well as rates of persistence and clearance in a subset of EBC pts from monarchE using the clinically validated SignateraTM ctDNA assay. Methods Samples were analyzed from a selected subset of pts (n=178; 84 from abemaciclib+ET arm; 94 ET alone) who had blood collected both before initiating protocol directed therapy, Visit 1 (V1; randomization + ≤ 3 days) and near completion of the 2-year treatment period, Visit 27 (V27; 24 months +/- 5 days from V1). This cohort of pts was enriched for overall IDFS events compared to the total monarchE study population, however pts with relapses occurring prior to V27 were excluded. Primary tumors from selected pts were subjected to whole exome sequencing (WES). Selected samples included those from pts with a range of tumor mutation burden. Cell-free DNA was extracted from 356 plasma samples. A Signatera ctDNA assay was developed for each pt based on up to 16 variants detected by WES from each baseline tumor sample. Results The IDFS event rate for the selected subset (n=178) was 39.3% (abemaciclib+ET 34.5% [29/84] and ET alone 43.6% [41/94]). Ten pts (5.6%) were initially ctDNA+ and 42 pts (23.6%) were persistently (7 pts, 3.9%) or became (35 pts, 19.7%) ctDNA+ at V27. Notably, 70% of pts who were initially ctDNA+ and 100% of pts who were either persistently ctDNA+ or became ctDNA+ experienced recurrence (see Table 1). In contrast, none of the 3 pts that cleared ctDNA+ developed recurrence. In pts persistently ctDNA negative (neg), 28 (21.1%) experienced recurrence. Overall, 10% (7/70) of pts with IDFS events were initially ctDNA+ and 50% (35/70) had detectable ctDNA at V27. Conclusions Detection of ctDNA soon after completing neoadjuvant chemotherapy was infrequent, but also associated with a high-risk of recurrence. ctDNA positivity was common at the end of the 2-year treatment period and was highly prognostic with all pts subsequently developing disease recurrence. Importantly, ~30% of pts with early ctDNA detection ultimately dropped below detection limits during the 2-year treatment period and none developed recurrence. Although some pts remained persistently ctDNA neg during this study and experienced recurrence, the delay in recurrence may indicate longitudinal benefit from remaining ctDNA neg. Future planned analysis of an expanded cohort reflective of the intent to treat population including additional timepoints within the 2-year treatment period will further define how ctDNA dynamics may identify pts at high-risk of recurrence. Table 1. Summary of Subset Positivity Rates. + positive; - negative; NA, not applicable. Citation Format: Sherene Loi, Stephen Johnston, Carlos Arteaga, Stephanie Graff, Sarat Chandarlapaty, Matthew Goetz, Christine Desmedt, Jorge Reis-Filho, Hironobu Sasano, Vanessa Rodrik-Outmezguine, Anthony Sireci, Cynthia Sandoval Rubenstein, Helen Won, Deli Liu, Lacey M. Litchfield, Nicholas Turner. Results from a pilot study exploring ctDNA detection using a tumor-informed assay in the monarchE trial of adjuvant abemaciclib with endocrine therapy in HR+, HER2-, node-positive, high-risk early breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PS06-01.

Abstract Background Two years of adjuvant abemaciclib combined with endocrine therapy (ET) resulted in significant and clinically meaningful improvement in invasive disease-free survival (IDFS) and distant relapse-free survival (DRFS) in patients (pts) with HR+, HER2-, node-positive, high-risk early breast cancer in the monarchE trial (NCT03155997). Abemaciclib benefit was sustained beyond the completion of treatment (tx) with deepening magnitude of absolute benefit in IDFS and DRFS at 5 years. Here, we evaluate comprehensive molecular profiling of archived primary tumor tissue and association with clinical outcomes. Methods For biomarker analysis, a proportionally stratified random sampling case-cohort design was utilized to include all patients who experienced an IDFS event at a pre-specified interim analysis with a median follow up of 54 months. A cohort of 895 pts (189 with IDFS event) in the abemaciclib + ET arm was matched 1:1 with 903 pts (270 with IDFS event) in the ET arm. Baseline primary tumor samples underwent exome-capture RNA sequencing (RNAseq; n=1324, 23% intent-to-treat (ITT) population) and paired tumor-normal (germline blood samples) whole-exome sequencing (WES; n=1234, 22% ITT population). Expression-based intrinsic subtypes (i.e., luminal A (LumA), luminal B (LumB), HER2-enriched (HER2E), basal- and normal-like) were characterized using the Absolute Intrinsic Molecular Subtyping model. The 21-gene expression signature score (Oncotype DX test) was inferred from RNAseq; samples were categorized into lower (0-25) and high (26-100) risk groups. To investigate associations of biomarkers with abemaciclib benefit, WES genomic events including oncogenic and hotspot mutations by OncoKB, and copy number events, of incidence &gt;9% were pre-selected. Results The biomarker subset of monarchE was reflective of the ITT population. A total of 1190 tumors (abemaciclib+ET n= 605; ET alone n=585) yielded adequate RNAseq results. Intrinsic subtype distribution was consistent across tx arms (Table A). Low tumor purity limited assessment of the normal-like subtype. The 4-year IDFS benefit of abemaciclib was consistent across all subtypes. LumA cancers had the lowest risk of recurrence while HER2E and basal-like subtypes had the highest. Inferred 21-gene expression signature score showed similar benefits from abemaciclib in both lower and high-risk groups (Table B). A total of 1173 tumors yielded adequate WES results (abemaciclib+ET n=580; ET alone n=593). Consistent abemaciclib benefit was observed across the most frequently altered genes (Table C). In exploratory analysis, lower benefit from abemaciclib was seen in the subset of focal-high level MYC amplified tumors (n=176, HR 1.30, 95% CI, 0.77, 2.20) compared to MYC non-amplified tumors (n=997, HR 0.62, 95% CI, 0.47, 0.80, nominal interaction p=0.014). The treatment benefit of abemaciclib was observed across all subpopulations of altered genes based on gene expression data. Conclusions Adjuvant abemaciclib+ET maintained IDFS benefit compared to ET alone across all molecular subtypes as measured by RNAseq. Benefit was consistent across most altered genes assessed by WES, except for the subset of tumors with MYC amplification. Additional research is necessary to confirm these findings. Table. Citation Format: Nicholas Turner, Jorge Reis-Filho, Matthew Goetz, Christine Desmedt, Sarat Chandarlapaty, Hironobu Sasano, Carlos Arteaga, Sherene Loi, Stephanie Graff, Deli Liu, Vanessa Rodrik-Outmezguine, Anthony Sireci, Cynthia Sandoval Rubenstein, Helen Won, Lacey M. Litchfield, Maria Munoz, Stephen Johnston. Genomic and transcriptomic profiling of primary tumors from patients with HR+, HER2-, node-positive, high-risk early breast cancer in the monarchE trial [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr GS03-06.