Matched molecular pairs (MMPs) capture the local chemical edits that medicinal chemists routinely use to design analogs, but existing ML approaches either operate at the whole-molecule level with limited edit controllability or learn MMP-style edits from restricted settings and small models. We propose a variable-to-variable formulation of analog generation and train a foundation model on large-scale MMP transformations (MMPTs) to generate diverse variables conditioned on an input variable. To enable practical control, we develop prompting mechanisms that let the users specify preferred transformation patterns during generation. We further introduce MMPT-RAG, a retrieval-augmented framework that uses external reference analogs as contextual guidance to steer generation and generalize from project-specific series. Experiments on general chemical corpora and patent-specific datasets demonstrate improved diversity, novelty, and controllability, and show that our method recovers realistic analog structures in practical discovery scenarios.
Small language models (SLMs) in the 100M-10B parameter range increasingly power production systems, yet whether they possess the internal emotion representations recently discovered in frontier models remains unknown. We present the first comparative analysis of emotion vector extraction methods for SLMs, evaluating 9 models across 5 architectural families (GPT-2, Gemma, Qwen, Llama, Mistral) using 20 emotions and two extraction methods (generation-based and comprehension-based). Generation-based extraction produces statistically superior emotion separation (Mann-Whitney p = 0.007; Cohen's d = -107.5), with the advantage modulated by instruction tuning and architecture. Emotion representations localize at middle transformer layers (~50% depth), following a U-shaped curve that is architecture-invariant from 124M to 3B parameters. We validate these findings against representational anisotropy baselines across 4 models and confirm causal behavioral effects through steering experiments, independently verified by an external emotion classifier (92% success rate, 37/40 scenarios). Steering reveals three regimes -- surgical (coherent text transformation), repetitive collapse, and explosive (text degradation) -- quantified by perplexity ratios and separated by model architecture rather than scale. We document cross-lingual emotion entanglement in Qwen, where steering activates semantically aligned Chinese tokens that RLHF does not suppress, raising safety concerns for multilingual deployment. This work provides methodological guidelines for emotion research on open-weight models and contributes to the Model Medicine series by bridging external behavioral profiling with internal representational analysis.
Celina DeBiasio, Jillian Dhawan, Quentin Nakonechny
et al.
Leiomyosarcoma (LMS) of the skin is a rare malignant neoplasm that can have varied clinical presentations, even mimicking benign lesions, making diagnosis challenging. We describe a 74-year-old male with a history of subcutaneous LMS of the left leg treated successfully with surgery, chemotherapy, and radiation over 10 years prior, who presented with an incidental papule on the lower back resembling an accessory nipple. Biopsy revealed a spindle cell tumour with diffuse smooth muscle actin and desmin positivity and a high Ki-67 index, consistent with a non-metastatic subcutaneous LMS. Wide local excision confirmed dermal and subcutaneous involvement with negative margins. This case is notable for the occurrence of two distinct primary LMS lesions more than a decade apart, underscoring the importance of long-term surveillance in high-risk patients. Given the potential for recurrence or metastasis, careful histopathologic evaluation and ongoing monitoring remain critical in the management of LMS.
In this article, we propose a novel approach for plant hierarchical taxonomy classification by posing the problem as an open class problem. It is observed that existing methods for medicinal plant classification often fail to perform hierarchical classification and accurately identifying unknown species, limiting their effectiveness in comprehensive plant taxonomy classification. Thus we address the problem of unknown species classification by assigning it best hierarchical labels. We propose a novel method, which integrates DenseNet121, Multi-Scale Self-Attention (MSSA) and cascaded classifiers for hierarchical classification. The approach systematically categorizes medicinal plants at multiple taxonomic levels, from phylum to species, ensuring detailed and precise classification. Using multi scale space attention, the model captures both local and global contextual information from the images, improving the distinction between similar species and the identification of new ones. It uses attention scores to focus on important features across multiple scales. The proposed method provides a solution for hierarchical classification, showcasing superior performance in identifying both known and unknown species. The model was tested on two state-of-art datasets with and without background artifacts and so that it can be deployed to tackle real word application. We used unknown species for testing our model. For unknown species the model achieved an average accuracy of 83.36%, 78.30%, 60.34% and 43.32% for predicting correct phylum, class, order and family respectively. Our proposed model size is almost four times less than the existing state of the art methods making it easily deploy able in real world application.
Sebastian Holm, MD, Mario Zambrana, MD, Juan E. Berner, MD, PhD
et al.
Summary:. Generative artificial intelligence (AI) large language models are an emerging technology, with ChatGPT and Gemini being 2 well-known examples. The current literature discusses clinical applications and limitations of AI, but its role in research has not yet been extensively evaluated. This study aimed to assess the role of ChatGPT and Gemini in developing novel and clinically relevant research ideas (RIs) for systematic reviews (SRs) in head and neck reconstruction. ChatGPT and Gemini were prompted to provide 10 novel and clinically relevant RIs for SRs in the following domains: head and neck reconstruction in general, microsurgery, and complications in reconstructive head and neck procedures. A comprehensive search was then performed for SRs in MEDLINE, Cochrane Library, and Embase to determine the novelty of the RIs generated. A total of 60 RIs were generated, with half created by ChatGPT and the other half by Gemini. Overall, 3613 entries were found through the literature search. After deduplication and screening, a total of 50 studies that partially addressed the AI-generated RIs were identified and were included in the present review. Out of the 60 AI-generated RIs, 42 had not been previously studied and were therefore considered novel. No statistically significant differences were found between the outputs generated by Gemini and ChatGPT. Both ChatGPT and Gemini were able to effectively generate novel and clinically relevant RIs for SRs, although their suggestions were generally broad. This study demonstrated that AI could potentially aid in the process of conducting novel SRs.
BackgroundSpinal diseases related pain represents a critical clinical issue that demands urgent resolution. Current treatment and assessment strategies predominantly focus on peripheral mechanisms. The application of functional magnetic resonance imaging (fMRI) offers a promising approach to identifying potential central targets for intervention.MethodsWe retrospectively included 31 patients with spinal diseases related pain and 32 controls with non-spinal, orthopedic complaints (no chronic neurological or psychiatric disorders). All participants underwent resting-state brain fMRI (eyes closed, awake). We quantified amplitude of low-frequency fluctuations (ALFF) with mean normalization (mALFF) and z-transformation (zALFF), regional homogeneity (ReHo; 27-voxel neighborhood), seed-based functional connectivity (FC; pre/postcentral seeds), and degree centrality (DC; binary and weighted). Between group tests used voxel-wise two-sample t_tests with Gaussian random field (GRF) correction.ResultsPatient group was associated with increased m/zALFF in right cerebellar lobule IX and right Superior Frontal Gyrus, medial part, and lower activity in bilateral postcentral gyri and the cuneus, decreased m/zALFF in bilateral postcentral gyri. ReHo analysis confirmed reduced local synchrony in postcentral regions, spatially overlapping with ALFF findings. FC analyses revealed enhanced cerebellar-thalamic connectivity (Crus1/2, thalamus) but reduced connectivity in sensorimotor and higher-order cortical networks. DC showed hyperconnectivity in left cerebellar Crus I with reduced Superior Frontal Orbital (Frontal_Sup_Orb). All findings survived GRF correction at the pre_specified thresholds.ConclusionResting-state brain fMRI indicates a cerebello-thalamo-cortical alteration pattern in spinal diseases related pain featuring cerebellar involvement, prefrontal subspecialization, and multilevel sensorimotor disruption. These cross-sectional associations may inform hypothesis-generation for future neuromodulation studies and provide candidate biomarkers for monitoring, pending prospective validation.
Background With increasing life expectancy, there has been a progressive rise in the proportion of patients (pts) aged ≥80 years diagnosed with myeloproliferative neoplasms (MPNs). However, data specifically focusing on very elderly pts with polycythemia vera (PV) remain limited. Aim: To describe the clinical characteristics and disease course of PV in a real-life cohort of pts aged ≥80 years. Methods: Between 1/2000 and 06/2024, 125 consecutive pts aged ≥80 years were diagnosed with PV across nine hematology centers. These pts were enrolled in the retrospective and prospective databases of the Latial group for Ph-negative MPNs. Diagnoses were reviewed according to WHO 2022 criteria. Results: Main characteristics at diagnosis are shown in the Table. Bone marrow (BM) biopsy was performed in 18 of 114 evaluable pts (15.8%). Median JAK2 V617F allele burden was 38.8% (IQR 11.4–65). At diagnosis, symptoms were reported in 24 of 81 evaluable pts (29.6%), with pruritus present in 10 (41.6%). A history of thrombotic events was found in 33 pts (26.4%). Hydroxyurea (HU) was initiated in 121 pts, typically within one month from diagnosis (IQR 0.1–3.8). HU was discontinued in 23 pts (18.8%), mainly due to intolerance (16/23). Of these, only 3 pts (13%) received ruxolitinib as 2nd line therapy, while 13 (56.6%) received no further cytoreductive treatment. During follow-up, thrombotic events occurred in 11 pts (8.8%), and disease progression to fibrotic or blastic phase was observed in 4 cases. At the last follow-up, 34 pts had died, 33 were lost to follow-up, 58 were alive. The 60- and 120-month cumulative overall survival (OS) rates were 71% (95%CI 82.2–59.8) and 31.9% (95%CI 49.2–14.6), respectively. Conclusion: This real-life cohort study of very elderly PV patients reveals key insights into current clinical practice. BM biopsy was infrequently performed, and nearly all pts received HU promptly after diagnosis. However, there was a noticeable hesitancy in initiating 2nd line therapy with ruxolitinib following HU discontinuation. The high rate of pts lost to follow-up, likely due to challenges in accessing care, limits the robustness of survival analysis. Nevertheless, the observed OS was comparable to that of the general population in the same age group.
Abstract E2F family genes are common transcription factors, abnormal in several malignant tumors. However, their complex involvement in colorectal cancer, particularly in prognosis, immune infiltration, and mutational landscape, remains unclear. We conducted a study using gene expression data from the TCGA and GEO datasets to examine the abnormal expression of E2Fs in colorectal cancer. And we performed consensus clustering and differential gene expression analyses to identify E2Fs-related genes. Then, we used Lasso regression and multivariate Cox regression to create a prognostic risk model for colorectal cancer. We analyzed the differences between the E2Fs-based gene risk and various clinical characteristics, gene mutations, immune cell infiltration, immunotherapy responses, and drug sensitivity using clinicopathological data, single-cell RNA sequences, multiple immune algorithms. Finally, we have developed a prognostic risk model that includes FMO5, NDUFA11, LIPG, FIGNL1, MOGAT2, and GZMB. We observed significant differences in clinical characteristics, immune cell infiltration, gene mutation landscapes, immunotherapy responses, and drug sensitivity between the high-risk and low-risk groups. The novel E2Fs-based gene risk model shows significant potential for contributing to the evaluation of prognosis and predicting immunotherapeutic outcomes for colorectal cancer patients.
Neoplasms. Tumors. Oncology. Including cancer and carcinogens
Mendelian randomization uses genetic variants as instrumental variables to make causal inferences about the effects of modifiable risk factors on diseases from observational data. One of the major challenges in Mendelian randomization is that many genetic variants are only modestly or even weakly associated with the risk factor of interest, a setting known as many weak instruments. Many existing methods, such as the popular inverse-variance weighted (IVW) method, could be biased when the instrument strength is weak. To address this issue, the debiased IVW (dIVW) estimator, which is shown to be robust to many weak instruments, was recently proposed. However, this estimator still has non-ignorable bias when the effective sample size is small. In this paper, we propose a modified debiased IVW (mdIVW) estimator by multiplying a modification factor to the original dIVW estimator. After this simple correction, we show that the bias of the mdIVW estimator converges to zero at a faster rate than that of the dIVW estimator under some regularity conditions. Moreover, the mdIVW estimator has smaller variance than the dIVW estimator.We further extend the proposed method to account for the presence of instrumental variable selection and balanced horizontal pleiotropy. We demonstrate the improvement of the mdIVW estimator over the dIVW estimator through extensive simulation studies and real data analysis.
Daniel Wolf, Tristan Payer, Catharina Silvia Lisson
et al.
Self-supervised pre-training of deep learning models with contrastive learning is a widely used technique in image analysis. Current findings indicate a strong potential for contrastive pre-training on medical images. However, further research is necessary to incorporate the particular characteristics of these images. We hypothesize that the similarity of medical images hinders the success of contrastive learning in the medical imaging domain. To this end, we investigate different strategies based on deep embedding, information theory, and hashing in order to identify and reduce redundancy in medical pre-training datasets. The effect of these different reduction strategies on contrastive learning is evaluated on two pre-training datasets and several downstream classification tasks. In all of our experiments, dataset reduction leads to a considerable performance gain in downstream tasks, e.g., an AUC score improvement from 0.78 to 0.83 for the COVID CT Classification Grand Challenge, 0.97 to 0.98 for the OrganSMNIST Classification Challenge and 0.73 to 0.83 for a brain hemorrhage classification task. Furthermore, pre-training is up to nine times faster due to the dataset reduction. In conclusion, the proposed approach highlights the importance of dataset quality and provides a transferable approach to improve contrastive pre-training for classification downstream tasks on medical images.
Olivia A. C. Lamers, Meta Roestenberg, Martine C. de Vries
et al.
Abstract Background Scientific advancement, including the testing and licensing of new drugs, relies heavily on clinical trials with healthy individuals. The motivations of clinical trial participants have been discussed intensively, as some worry that financial compensation may distract from the intrinsic risk of clinical research. Herein, we investigated the motivations and decisional factors influencing SARS-CoV-2 clinical trial participants. Moreover, since most surveys are administered after clinical trial participation, we were interested in whether the results were tainted by recall bias. Methods This was a cross-sectional observational study. Participants were administered a survey on two occasions, once before and once after participation in a clinical trial. The primary outcomes were the motivations and decisional factors of SARS-CoV-2 vaccine trial participants and the difference between the surveys collected before and after clinical trial participation. Results The survey response rate was 149/200 (75%). SARS-CoV-2 vaccine trial participants were mostly motivated by the desire to contribute to science and help others. Answers collected before and after the trial were not statistically different, indicating the absence of recall bias. Conclusion The decision-making process of clinical trial participants is complex and multi-faceted. Previous studies have shown that clinical trial participants have mixed motivations but never to the extent reported in the current survey. Here, we present a theoretical framework that attempts to explain how different motivational factors may contribute to decision forming.
Navya Jaiswal, Shrijeet Chakraborti, Neema Tiwari
et al.
Introduction: Peripheral neuropathy is common in clinical practice, with a reported prevalence of 2.4% in the general population. There are numerous aetiologies for peripheral neuropathy like diabetes, ischaemia, vasculitis, inflammatory demyelinating disorders, nutritional deficiencies, paraproteinemic disorders, paraneoplastic syndromes, toxic exposures, and hereditary neuropathies. An exhaustive haematological, biochemical, and serological work-up, cerebrospinal fluid evaluation, electrodiagnostic tests, and nerve biopsy are required when overlapping clinical features present a diagnostic challenge.
Aim: To analyse the histopathological characteristics of nerve biopsies in individuals with peripheral neuropathy.
Materials and methods: This cross-sectional study was conducted in the Department of Pathology at Kasturba Medical College from January 2011 to December 2016. Nerve biopsies received in the Department of Pathology, Kasturba Medical College and Hospitals in the Ambedkar Circle and Attavar area, as well as at Government Wenlock Hospital, Mangaluru, Karnataka, India were studied. Clinical and laboratory data were collected from biopsy requisition forms and patient case records from the aforementioned hospitals. Results were presented in numbers and percentages.
Results: A total of 134 nerve biopsies were included in the study. The age range of all cases studied was 7-86 years, with a mean age of 51.8 years. The study population consisted of 63.4% males and 36.6% females, resulting in a male-to-female ratio of 1.7:1. Vasculitic Neuropathy (VN) accounted for 38.1% cases, followed by chronic inflammatory neuropathies (21.7%) and Diabetic Neuropathy (DN) (12.6%). Other diagnosis included ischaemic neuropathy (6.7%), Hereditary Motor and Sensory Neuropathy (HMSN) (3.7%), subacute inflammatory demyelinating neuropathy (3.0%), as well as two cases each of Hansen’s neuritis, amyloid neuropathy, and acute inflammatory demyelinating neuropathy (1.5% each). One case of toxic neuropathy (0.7%) was identified, while 9.0% of cases displayed histological features that were either non specific or not characteristic of any specific diagnosis.
Conclusion: Vasculitic neuropathy was found to be the most common aetiology of peripheral neuropathy, followed by Chronic Inflammatory Demyelinating Polyneuropathy (CIDP) and DN. The histopathological examination of nerve biopsies is a useful tool in distinguishing between diseases with overlapping clinical features, confirming or excluding vasculitis, and diagnosing hereditary conditions in settings where genetic testing is unavailable or unaffordable.
Militsa Rakina, Irina Larionova, Julia Kzhyshkowska
M1/M2 paradigm of macrophage plasticity has existed for decades. Now it becomes clear that this dichotomy doesn't adequately reflect the diversity of macrophage phenotypes in tumor microenvironment (TME). Tumor-associated macrophages (TAMs) are a major population of innate immune cells in the TME that promotes tumor cell proliferation, angiogenesis and lymphangiogenesis, invasion and metastatic niche formation, as well as response to anti-tumor therapy. However, the fundamental restriction in therapeutic TAM targeting is the limited knowledge about the specific TAM states in distinct human cancer types. Here we summarized the results of the most recent studies that use advanced technologies (e.g. single-cell RNA sequencing and spatial transcriptomics) allowing to decipher novel functional subsets of TAMs in numerous human cancers. The transcriptomic profiles of these TAM subsets and their clinical significance were described. We emphasized the characteristics of specific TAM subpopulations – TREM2+, SPP1+, MARCO+, FOLR2+, SIGLEC1+, APOC1+, C1QC+, and others, which have been most extensively characterized in several cancers, and are associated with cancer prognosis. Spatial transcriptomics technologies defined specific spatial interactions between TAMs and other cell types, especially fibroblasts, in tumors. Spatial transcriptomics methods were also applied to identify markers of immunotherapy response, which are expressed by macrophages or in the macrophage-abundant regions. We highlighted the perspectives for novel techniques that utilize spatial and single cell resolution in investigating new ligand-receptor interactions for effective immunotherapy based on TAM-targeting.
Ralph P. Eatough, Gregory Desvignes, Kuo Liu
et al.
BlackHoleCam is a project funded by a European Research Council Synergy Grant to build a complete astrophysical description of nearby supermassive black holes by using a combination of radio imaging, pulsar observations, stellar astrometry and general relativistic magneto-hydrodynamic models. BlackHoleCam scientists are active partners of the Event Horizon Telescope Consortium. In this talk I will discuss the use of pulsars orbiting Sagittarius A* for tests of General Relativity, the current difficulties in detecting such sources, recent results from the Galactic Centre magnetar PSR J1745-2900 and how BlackHoleCam aims to search for undiscovered pulsars in the Galactic Centre.
The cytosolic dipeptidyl-aminopeptidases 8 (DPP8) and 9 (DPP9) belong to the DPPIV serine proteases with the unique characteristic of cleaving off a dipeptide post-proline from the <i>N</i>-termini of substrates. To study the role of DPP8 and DPP9 in breast cancer, MCF-7 cells (luminal A-type breast cancer) and MDA.MB-231 cells (basal-like breast cancer) were used. The inhibition of DPP8/9 by 1G244 increased the number of lysosomes in both cell lines. This phenotype was more pronounced in MCF-7 cells, in which we observed a separation of autophagosomes and lysosomes in the cytosol upon DPP8/9 inhibition. Likewise, the shRNA-mediated knockdown of either DPP8 or DPP9 induced autophagy and increased lysosomes. DPP8/9 inhibition as well as the knockdown of the DPPs reduced the cell survival and proliferation of MCF-7 cells. Additional treatment of MCF-7 cells with tamoxifen, a selective estrogen receptor modulator (SERM) used to treat patients with luminal breast tumors, further decreased survival and proliferation, as well as increased cell death. In summary, both DPP8 and DPP9 activities confine macroautophagy in breast cancer cells. Thus, their inhibition or knockdown reduces cell viability and sensitizes luminal breast cancer cells to tamoxifen treatment.
Clustering bipartite graphs is a fundamental task in network analysis. In the high-dimensional regime where the number of rows $n_1$ and the number of columns $n_2$ of the associated adjacency matrix are of different order, existing methods derived from the ones used for symmetric graphs can come with sub-optimal guarantees. Due to increasing number of applications for bipartite graphs in the high dimensional regime, it is of fundamental importance to design optimal algorithms for this setting. The recent work of Ndaoud et al. (2022) improves the existing upper-bound for the misclustering rate in the special case where the columns (resp. rows) can be partitioned into $L = 2$ (resp. $K = 2$) communities. Unfortunately, their algorithm cannot be extended to the more general setting where $K \neq L \geq 2$. We overcome this limitation by introducing a new algorithm based on the power method. We derive conditions for exact recovery in the general setting where $K \neq L \geq 2$, and show that it recovers the result in Ndaoud et al. (2022). We also derive a minimax lower bound on the misclustering error when $K = L$ under a symmetric version of our model, which matches the corresponding upper bound up to a factor depending on $K$.
Claire E. Halloran, Filiberto Fele, Malcolm D. McCulloch
This paper investigates how spatiotemporal heterogeneity in inflexible residential heat pump loads affects the need for storage and generation in the electricity system under business-as-usual and low-carbon emissions budgets. Homogeneous and heterogeneous heat pump loads are generated using population-weighted average and local temperature, respectively, assuming complete residential heat pump penetration. The results of a storage and generation optimal expansion model with network effects for spatiotemporally homogeneous and heterogeneous load profiles are compared. A case study is performed using a 3-bus network of London, Manchester, and Glasgow in Britain for load and weather data for representative weeks. Using heterogeneous heating demand data changes storage sizing: under a business-as-usual budget, 26% more total storage is built on an energy and power basis, and this storage is distributed among all of the buses in the heterogeneous case. Under a low-carbon budget, total energy storage at all buses increases 2 times on an energy basis and 40% on a power basis. The energy to power ratio of storage at each bus also increases when accounting for heterogeneity; this change suggests that storage will be needed to provide energy support in addition to power support for electric heating in high-renewable power systems. Accounting for heterogeneity also increases modeled systems costs, particularly capital costs, because of the need for higher generation capacity in the largest load center and coincidence of local peak demand at different buses. These results show the importance of accounting for heat pump load heterogeneity in power system planning.
Objective:Pediatric cardiac catheterization and angiography are two of the methods used in the diagnosis and treatment of patients with congenital heart disease. Anesthesia approaches in these patients are special and come with many anesthetic challenges. In this framework, the objective of this study is to evaluate our anesthesia experience and complications in catheterization procedures performed in the pediatric angiography laboratory.Method:This study was conducted with patients who underwent diagnostic or interventional catheterization in the pediatric angiography laboratory, between August 1st, 2020, and December 31st, 2021. Demographic and clinical characteristics of these patients, including their cardiac diagnosis, gender, weight, procedural characteristics, and anesthesia management principles employed during the procedure and complications, were evaluated.Results:A total of 390 procedures were applied to 360 patients during the period covered by the study. The median age and weight of these patients were three months [interquartile range (IQR) 20 days-7 years)] and 7 kg (IQR 3.4-24), respectively. Of these patients, 51% were male, and 49% were female. Of the 390 procedures, 134 were performed diagnostically, and 256 were performed invasively. The median duration of the procedure was 35 minutes (IQR 25-60). The catheterization procedure was performed under general anesthesia in 33% of the cases. A total of 52 (13.3%) complications were observed during the procedures. During the anesthesia management, hypotension was observed in 17 patients, desaturation in 10 patients, and rhythm disturbances in 7 patients. Two patients needed extracorporeal membrane oxygenation support during the procedure. Four patients had to be operated on urgently. No patient was lost due to the procedure.Conclusion:Anesthesia management is characteristic during the catheterization procedure in pediatric cases with congenital heart disease. A case-specific anesthesia approach should be preferred, taking into account factors such as the type of cardiac pathology, hemodynamic characteristics, and type of procedure.
<i>Taraxacum officinale</i> (dandelion) is often used in traditional Chinese medicine for the treatment of cancer; however, the downstream regulatory genes and signaling pathways mediating its effects on breast cancer remain unclear. The present study aimed to explore the effects of luteolin, the main biologically active compound of <i>T. officinale</i>, on gene expression profiles in MDA-MB-231 and MCF-7 breast cancer cells. The results revealed that luteolin effectively inhibited the proliferation and motility of the MDA-MB-231 and MCF-7 cells. The mRNA expression profiles were determined using gene expression array analysis and analyzed using a bioinformatics approach. A total of 41 differentially expressed genes (DEGs) were found in the luteolin-treated MDA-MB-231 and MCF-7 cells. A Gene Ontology analysis revealed that the DEGs, including AP2B1, APP, GPNMB and DLST, mainly functioned as oncogenes. The human protein atlas database also found that AP2B1, APP, GPNMB and DLST were highly expressed in breast cancer and that AP2B1 (cut-off value, 75%) was significantly associated with survival rate (<i>p</i> = 0.044). In addition, a Kyoto Encyclopedia of Genes and Genomes pathway analysis revealed that the DEGs were involved in T-cell leukemia virus 1 infection and differentiation. On the whole, the findings of the present study provide a scientific basis that may be used to evaluate the potential benefits of luteolin in human breast cancer. Further studies are required, however, to fully elucidate the role of the related molecular pathways.