Sorbonne University

We investigate the classical aspects of quantum theory and under which description quantum theory does appear classical. Although such descriptions or variables are known as “ontological” or “hidden,” they are not hidden at all but are dual classical states (in the sense of the general classical–quantum duality of nature). We analyze and interpret the dynamical scenario in an inherent quantum structure: (i) We show that the use of the known φ states in the circle [F. London, Z. Phys. 37, 915 (1926) and G. ’t Hooft, “The hidden ontological variable in quantum harmonic oscillators,” arXiv 2407.18153 (2024)] takes a true dimension only when the system is subjected to the minimal group representation action of the metaplectic group Mp(n). The Mp(n) Hermitian structure fully covers the symplectic Sp(n) group and, in certain cases, OSp(n). (ii) We compare the circle φ states and the cylinder ξ states in configuration space with the two sectors of the full Mp(2) Hilbert space corresponding to the even and odd n harmonic oscillators and their total sum. (iii) We compute the projections of the Mp(2) states on the circle φ and cylinder ξ states. The known London circle states are not normalizable. We compute here the general coset coherent states α,φ in the circle, with α being the coherent complex parameter. It allows full normalizability of the complete set of the circle states. (iv) The London states (ontological in ’t Hooft’s description) completely classicalize the inherent quantum structure only under the action of the Mp(n) minimal group representation. (v) For the coherent states in the cylinder (configuration space), all functions are analytic in the disk z=ωe−iφ<1. For the general coset coherent states α,φ in the circle, the complex variable is z′=ze−iα*/2: The analytic function is modified by the complex phase (φ − α*/2). (vi) The analyticity z′=ze−Imα/2<1 occurs when Im α ≠ 0 because of normalizability and Im α > 0 because of the identity condition. The circle topology induced by the α,φ coset coherent state also modifies the ratio of the disk due to the displacement by the coset. (vii) For the coset coherent cylinder states in configuration space, the classicalization is stronger due to screening exponential factors e−2n2, e−(2n+1/2), and e−(2n+1/2)2 for large n arising in the Mp(2) projections on them. The generalized Wigner function shows a bell-shaped distribution and stronger classicalization than the square norm functions. The application of the minimal group representation immediately classicalizes the system, with Mp(2) emerging as the group of the classical–quantum duality symmetry.

This book is devoted on some recent investigations of some classes partial differential equations in Sobolev and analytic spaces. The book contains twelve chapters. Chapter 1 is entirely devoted to the presentation of definitions and results necessary as a result of this work. We first recall a few basic results on the linear, metric, normed and Banach Spaces and its properties. These are used in particular to introduce the different concepts of weak solutions to PDEs. Chapter 2 is titled ”Lebesgue Integration”, it is devoted to the study of measure and integration, Lebesgue measurable functions and general measure spaces, where there are many proved results. The purpose of Chapter 3 is to present results according to the Lp spaces, which contains definitions, separability, duality and general Lp spaces with its norms. Elements of theory of distributions are introduced in Chapter 4. It is introduced the Fourier transform in the spaces S and S0 and they are deducted some important properties. Sobolev spaces and Bourgain spaces are discussed in Chapter 5. Chapter 6 is devoted on the original method for KdV equations in Sobolev spaces. In Chapter 7 are investigated some fifth order shallow water equations for local and global well posedness. Higher order dispersive equations are studied in Chapter 8. They are given criteria for existence of local global solutions in some analytic spaces. In Chapter 9 are investigated some classes fifth order Kadomtsev-Petviashvili I equations in Bourgain spaces for existence of solutions, continuous dependence on initial data and Gevrey regularity in t. They are studied some classes KadomtsevPetviashvili II equations in anisotropic Gevrey spaces. Chapter 10 is devoted on the generalized Kadomtsev-Petviashvili I equation. Coupled systems of KdV equations are introduced and studied in Chapter 11. In Chapter 12 are discussed coupled systems of generalized KdV equations.

The humanist theory of the nude is one of the places where what can be called a ‘poor metaphysics’ developed during the Renaissance. To construct the concept of the nude as a representation of man in his own right, art theorists used common scholastic categories such as substance and accident, form and matter, potentiality and actuality, quantity and quality, whole and part, soul and body. Resolutely poor in its object – the human body, the work of art – and in its form – technical treatise, fictional dialogue, or simple working notes – this reflection is nonetheless rich and original because of what constitutes its very weakness: the contamination of the Aristotelian metaphysical tradition with Neoplatonism, Vitruvianism, elements of natural philosophy, musical theory, and even Kabbalah. It testifies less to the permanence of scholastic metaphysics during the Renaissance than to the ingenious adaptation of its tools to new, humbler, and more rebellious objects of thought.

Importance Preventive efforts in pregnancy-related alloimmunization have considerably decreased the prevalence of hemolytic disease of the fetus and newborn (HDFN). International studies are therefore essential to obtain a deeper understanding of the postnatal management and outcomes of HDFN. Taken together with numerous treatment options, large practice variations among centers may exist. Objectives To assess variations in postnatal management and outcomes of HDFN among international centers and to identify opportunities to improve care. Design, Setting, and Participants In this international, retrospective, cohort study, 31 expert centers from 22 countries retrieved data on neonates with HDFN managed between January 1, 2006, and July 1, 2021. Statistical analysis was performed from July 19, 2023, to October 28, 2024. Main Outcomes and Measures Main outcomes included the frequency of exchange transfusions, administration of intravenous immunoglobulin, administration of erythropoiesis-stimulating agents, and red blood cell transfusions, as well as the association of gestational age at birth with exchange transfusion frequency and risk factors for adverse neonatal outcomes. Results The study included 1855 neonates (median gestational age at birth, 36.4 weeks [IQR, 35.0-37.3 weeks]; 1034 boys [55.7%]), of whom 1017 (54.8%) received any form of antenatal treatment. Most neonates (1447 [78.0%]) had anti-D antibodies. Exchange transfusions were performed in 436 neonates (23.5%), with proportions in exchange transfusion frequency varying from 0% to 78% among centers. Intravenous immunoglobulin was administered to 429 of 1743 neonates (24.6%), with proportions varying from 0% to 100% among centers. A higher gestational age at birth was associated with a reduction in exchange transfusion frequency in neonates with intrauterine transfusion, decreasing from approximately 38.2% (13 of 34) at 34 weeks to 16.8% (18 of 107) after 37 weeks and 0 days. A weekly increase in gestational age at birth was associated with a 43.3% decrease (95% CI, 36.1%-49.7%) in the likelihood of adverse neonatal outcomes, and neonates who received an exchange transfusion were 1.55 (95% CI, 1.10-2.18) times more likely to experience unfavorable outcomes. Conclusions and Relevance In this cohort study of neonates with HDFN managed at 31 centers in 22 countries, significant practice variations in the postnatal management of HDFN were identified, highlighting the lack of, and need for, consensus. The study suggests that there is a potential beneficial clinical association of waiting for delivery until after 37 weeks and 0 days with frequency of exchange transfusions among neonates with HDFN. The framework to implement international guidelines is provided.

Background Less adequate cardiorespiratory fitness (CRF) is associated with several aspects of Alzheimer’s disease (AD) pathology, including neuroinflammation, neurodegeneration and synaptic dysfunction, all of which are known contributors to the clinical outcome – progressive cognitive decline [1]. AD‐associated biomolecular changes also seem to be attenuated in carriers of the functionally advantageous variant of the KLOTHO gene (KL‐VSHET) [2]. While KL‐VSHET and CRF both appear to mitigate aspects of AD pathology, they have been exclusively studied in isolation. Here we investigate whether the relationships between CRF (VO2 max) and cerebrospinal fluid (CSF) biomarkers of neurodegeneration, synaptic dysfunction, and inflammation differ for KL‐VSHET compared to non‐carriers (KL‐VSNC). Method The cohort, enriched for AD risk, consisted of cognitively unimpaired adults (N=132; MeanAGE=62.7) from the Wisconsin Registry for Alzheimer’s Prevention and Wisconsin Alzheimer’s Disease Research Center. Covariate‐adjusted (age, sex, parental AD history, APOE, and age difference between CSF sampling and exercise test) linear models examined the relationship between VO2 max and CSF biomarkers of neurodegeneration [α‐synuclein (α‐syn), neurofilament light polypeptide (NfL), total tau (tTau)], synaptic dysfunction [neurogranin (Ng)], and neuroinflammation [glial fibrillary acidic protein (GFAP), soluble triggering receptor expressed in myeloid cells (sTREM2), chitinase‐3‐like protein 1 (YKL‐40), interleukin 6 (IL‐6), S100 calcium‐binding protein B (S100B)] as a function of KLOTHO KL‐VS. Result The interaction between VO2 max and KL‐VSHET was significant for tTau (P=0.05), Ng (P=0.02), sTREM2 (P=0.02), and YKL‐40 (P=0.03) (Table 1; Figure 1), such that KL‐VSHET who were more fit had significantly lower levels of tTau, Ng, sTREM2, and YKL‐40 but not α‐syn, NfL, GFAP, IL‐6, or S100B (all Ps>0.63). Conclusion We report a synergistic relationship between KL‐VSHET and CRF with regard to neurodegeneration (tTau), synaptic dysfunction (Ng) and inflammation (sTREM2 and YKL‐40), suggesting a protective role for both KL‐VSHET and better cardiovascular fitness against unfavorable AD‐related changes. Their potentially shared biological mechanisms will require future investigations. References [1] Huuha, A.M., et al., Can exercise training teach us how to treat Alzheimer's disease? Ageing Res Rev, 2022. 75: p. 101559. [2] Driscoll, I., et al., AD‐associated CSF biomolecular changes are attenuated in KL‐VS heterozygotes. Alzheimers Dement (Amst), 2022. 14(1): p. e12383.

Background Glial fibrillary acidic protein (GFAP) is a marker of cerebral astrogliosis and occasionally elevated in patients with dementia. GFAP in cerebrospinal fluid (CSF), is routinely requested in referrals to neurochemistry laboratories; however, its ability to differentiate dementias and diagnostic capability is unclear. Our aim was to elucidate this, using two large datasets. Method First, GFAP data measured since 2015 was retrieved from the database of the Clinical Neurochemistry Laboratory at the Sahlgrenska University hospital. We then cross‐referenced with the Swedish dementia registry (SveDem). Here, information on ten different diagnoses such as early onset AD (EAD [<65 years]), late onset AD (LAD [≥65 years]), Parkinson disease with dementia (PDD), vascular dementia (VaD) and frontotemporal dementia (FTD), each with specific diagnostic criteria, were retrieved. The GFAP data was log10‐transformed, followed by an analysis of covariance (ANCOVA) and a subsequent post‐hoc Tukey’s test, with GFAP as dependent variable, diagnosis as independent variable and sex and age as covariates. Result In total, 1912 individuals (mean [SD] age, 71.9 [8.2] years; 52% male), were included. Lower log10‐transformed GFAP concentrations were seen in PDD (mean [SD], 2.68, [0.28] pg/mL), than in EAD, LAD, VaD and FTD; here, mean concentrations of 2.76 (0.24), 2.89 (0.23), 2.89 (0.32) and 2.76 (0.25) pg/mL were observed, respectively. In the post hoc analysis, GFAP differentiated VaD from EAD (p<0.001). PDD concentrations were significantly different from VaD (p<0.001) and LAD (p<0.001). Further, it also differentiated FTD from VaD (p=0.006) and LAD (p=0.001). Conclusion CSF GFAP could on a group level help differentiate VaD from EAD, FTD and PDD. Also, it could differentiate PDD from LAD. These results bear potential clinical relevance, where clinicians in some uncertain cases could use this marker as a differential tool.

Background The identification of biomarkers for Alzheimer's disease (AD) remains a significant challenge, particularly for the clinical severity of the disease. Recent studies have shown that plasma brain‐derived‐tau (BD‐Tau) could be a promising biomarker for the identification of AD‐type neurodegeneration. This study aimed to investigate the potential of BD‐Tau in differentiating various clinical stages of AD, ranging from cognitively unimpaired AD to severe dementia AD. Additionally, the study intended to examine the association of BD‐Tau with clinical severity and elucidate its dynamic behavior throughout the natural course of AD. Method Cross‐sectional and longitudinal EDTA plasma data, along with clinical information, were utilized from the Pitié‐Salpêtrière hospital INSIGHT (cognitively unimpaired individuals with amyloid PET) and SOCRATES cohorts (AD and non‐AD symptomatic neurodegenerative disease according to up‐to‐date international criteria). Plasma BD‐Tau was analyzed using the Gothenburg University homebrew Quanterix Simoa immunoassay. Result The results revealed that BD‐Tau is only elevated in individuals with symptomatic AD, whether the primary or secondary neurodegenerative disease, and not in other neurodegenerative conditions. Additionally, BD‐Tau follows the clinical dynamics of AD, as it is only elevated in symptomatic AD patients, while individuals with AD biological changes but no symptoms have normal levels of BD‐Tau. BD‐Tau levels were also found to be significantly correlated with episodic memory scores and global cognitive performance across a wide range of AD severity. Longitudinal analyses further support BD‐Tau's potential as an AD‐type neurodegeneration biomarker, as it was found to increase over time only in the symptomatic AD group and not in other groups. Conclusion These results support the notion that plasma BD‐Tau could be a promising specific biomarker for AD staging, as it follows the AD clinical severity spectrum and is not elevated in asymptomatic amyloidosis.

Background Social isolation (SI) and loneliness are related to several negative health outcomes, including cognitive decline and dementia. While both males and females experience increased SI as a function of age, studies have found that females experience greater loneliness than males, despite males reporting more physical isolation and smaller social networks. Females, independent of SI‐status, are also at increased risk of Alzheimer’s disease (AD). These sex‐related differences in SI and AD‐risk prompted us to examine whether sex and loneliness influenced the plasma AD biomarkers in a cohort of elderly individuals. Method Participants were enrolled in the Memory Education and Research Initiative (MERI) program. MERI participants completed a neuropsychological battery, clinical and psychiatric evaluation, and blood draw. Plasma Aβ40, Aβ42, and PTau231 concentrations were determined using single‐molecule array (SIMOA) platform. Plasma Aβ42/Aβ40 ratio was calculated. MERI participants were included if they gave blood, were cognitively‐normal defined by MMSE>27, age 50 years older, and completed the Profile of Mood States scale (POMS). POMS‐Loneliness item was coded to a dichotomous variable (0‐Not Lonely; 1‐Lonely). Result A total of 459 MERI participants (mean age=71 years; 59% females) were included. There were no significant differences between males and females for education, HAM‐D total score, or MMSE, only age. A 2x2 analysis of covariance (ANCOVA) with sex (male, female) and loneliness (not lonely, lonely), and age as a covariate, revealed a significant interaction between sex and loneliness with Aβ42 (p=0.014). Lonely females had significantly lower Aβ42 than males who are lonely. The same analysis revealed a significant main effect of sex on Aβ40 (p=0.014) and PTau231 (p=0.007); females had higher Aβ40 and lower PTau231, respectively. There were no significant main effects or interactions for Plasma ratio Aβ42/Aβ40. Conclusion Cognitively‐normal females who reported being lonely, as compared with males, had lower levels of plasma Aβ42 but not lower levels of plasma Aβ42/Aβ40 ratio, a better predictor of a positive amyloid PET scan. Future studies should examine if the results reflect increased brain amyloid deposition in lonely females. If so, interventions that mitigate loneliness may reduce overall risk for AD in females.

Background Cerebral pulsatility (PI) is reportedly higher in individuals with AD and MCI compared to age matched controls and has been associated with greater beta‐amyloid (Aß) burden, but its relationship to other neurodegenerative biomarkers is unknown. Higher cardiorespiratory fitness (CRF) positively affects vascular function and is associated with lower PI in several large cerebral vessels. The relationship between PI, CRF, and biomarkers for neurodegeneration have not yet been characterized. Our objective was to examine a potential CRF modification of the relationship between PI and cerebrospinal fluid (CSF) biomarkers of neurodegeneration, synaptic dysfunction, and neuroinflammation. Methods Cognitively unimpaired adults (n=32, Mean Age=64) from the Wisconsin Registry for Alzheimer's Prevention and the Wisconsin Alzheimer's Disease Research Center who met selection criteria were included. PI from bilateral MCA vessels were measured using 4D flow MRI and averaged (PI‐MCAAVG). CRF was measured as peak VO2PEAK during a graded exercise treadmill test. CSF was obtained via lumbar puncture and measured using the NeuroToolKit, a panel of robust prototype assays (Roche Diagnostics International Ltd). The toolkit included measures of: neurodegeneration [α‐synuclein (α‐syn) and neurofilament light polypeptide (NfL)], synaptic dysfunction [neurogranin (Ng), synaptosomal‐associated protein 5 (SNAP‐25), and neuronal activity related protein 2 (nptx2)], and neuroinflammation [glial fibrillary acidic protein (GFAP), soluble triggering receptor expressed in myeloid cells (sTREM2), and chitinase‐3‐like protein 1 (YKL‐40)]. A single multivariate regression model, adjusted for sex, age, and APOE4, examined whether the association between PI‐MCAAVG and CSF biomarkers varied as a function of CRF. Results There was a significant interaction between PI and VO2PEAK on α‐syn (ß = ‐22.66, p=0.04), Ng (ß =‐171.46, p=0.01), SNAP‐25 (ß = ‐0.35, p= 0.02), STREM2 (ß =‐1.04, p=0.02), and nptx2 (ß =‐1037, p=0.04) indicating more favorable profiles of CSF biomarkers with increasing VO2PEAK despite elevated PI. No significant interactions were found for NfL, GFAP, or YKL40 (all p’s > 0.05). Conclusion Our findings support the hypothesis that increased CRF influences the relationship between cerebral PI and CSF biomarkers related to neurodegeneration, synaptic dysfunction, and neuroinflammation indicating that improved CRF may offer protection against cerebrovascular alterations known to coincide with biomolecular changes in various types of dementia.

Background Increasing evidence supports the notion that vascular dysfunction contributes to the evolution of Alzheimer’s disease (AD). Cerebral pulsatility index (PI) is reportedly higher in AD and MCI compared to age matched controls and has been associated with greater beta‐amyloid (Aß) burden. Higher cardiorespiratory fitness (CRF) positively affects vascular function and is associated with lower PI in several large cerebral vessels. Our objective was to examine whether CRF modifies the relationship between PI, Aß burden, and core AD cerebrospinal fluid (CSF) biomarkers. Method Cognitively unimpaired adults (n=33, MeanAGE=64.0) from the Wisconsin Registry for Alzheimer's Prevention and the Wisconsin Alzheimer's Disease Research Center who met study criteria were included. Aß burden was measured as global cortical PiB‐PET. CSF biomarkers were measured with the NeuroTookKit, a panel of robust prototype assays (Roche Diagnostics International Ltd). PI from bilateral MCA vessels were measured using 4D flow MRI and averaged (PI‐MCAAVG). CRF was indexed as VO2peak during a graded exercise treadmill test. A single multivariate regression, (covariate‐adjusted for sex, age, and APOE4) examined whether CRF modifies the impact of PI‐MCAAVG on Aß burden and core‐AD CSF biomarkers [phosphorylated Tau 217 (pTau‐217), total Tau (tTau), Aß42/Aß40, pTau/Aß42]. When significant, the PI*CRF interaction term would indicate a differential effect of PI on Aß burden or CSF biomarker as a function of CRF. Result There was a significant interaction between PI and VO2peak on PET Aß burden (ß = ‐0.05, p= 0.04), Aß42/ Aß40 (ß=0.006, p=0.03), pTau (ß= ‐3.4, p=0.03), tTau (ß ‐29.8, p=0.04) and pTau/Aß42 (ß= ‐0.006, p=0.02), indicating less Aß burden and more favorable profiles for all core CSF biomarkers with increasing VO2peak despite elevated PI. Conclusion Our findings support the hypothesis that greater CRF modifies the effect of cerebral PI on PET Aß burden and core AD CSF biomarkers, indicating that improved VO2 seems to be protective against cerebrovascular alterations known to contribute to AD‐related biomolecular changes.

Background Recently, the development of ultra‐sensitive immunoassays has allowed for the detection, in blood, of proteins related to the pathophysiology of Alzheimer’s disease (AD), with phosphorylated tau (p‐tau) being the most promising. However, current methods are often limited by their ability to measure one analyte, lacking the potential for discovery and inclusion of additional biomarkers with supplemental value. In this pilot study, we explored proteomic changes using the novel NUcleic acid Linked Immuno‐Sandwich Assay (NULISA™) platform, focusing on patients with mild cognitive impairment (MCI). Method In this study, MCI patients with a confirmed Aβ status were recruited from two independent cohorts. A discovery cohort (mean[SD] age, 73.5 [5.5] years; 25 females [62.5%]) was classified by cerebrospinal fluid Aβ42/40 (Aβ+ =28; Aβ‐ =12). For the second cohort (mean[SD] age, 70.7 [7.31] years; 39 females[42.9%]; CDR 0.5), from the TRIAD study, amyloid positron emission tomography was utilized instead (Aβ+ =47; Aβ‐ =44). We performed the NULISAseq CNS Disease Panel measurements on the plasma samples (n=131) in two separate batch analyses. LIMMA models were used to evaluate differential expression of protein NPQ values between the two MCI (Aβ+ Vs. Aβ‐) patient groups, with a total of 121 proteins included in the analysis. Result In the discovery cohort (n=40), only p‐tau217 survived multiple comparison (Log2FC, 1.49; Padj <0.001). Further targets were present with a significant unadjusted p‐value (P <0.05) (Figure 1b). In the larger TRIAD cohort (n=91), p‐tau217 (Log2FC, 1.29; Padj <0.001), p‐tau231 (Log2FC, 0.76; Padj <0.05), p‐tau181 (Log2FC, 0.42; Padj <0.05) and GFAP (Log2FC, 0.78; Padj <0.05) remained significant after adjusting for multiple testing. Similarly to the discovery cohort, additional targets were identified as significantly changed with an unadjusted p‐value (P<0.05) (Figure 1A). This included PARK7, which was the only target present in both. Conclusion In this study, consisting of two independent groups of MCI patients characterized by Aβ burden, we utilized the NULISAseq CNS Disease Panel to identify dysregulated proteins of the prodromal stage of AD. This novel multiplex technology continues to demonstrate the importance of p‐tau217, p‐tau231, p‐tau181 and GFAP as indicators of cerebral amyloid deposition while offering additional markers that may increase their utility.

Background The immune complement system is key to the elimination of redundant neural connections in the brain through a process called synaptic pruning. In neurodegenerative diseases such as Alzheimer's disease (AD), this system may result in excessive synapse loss, leading to brain atrophy and cognitive impairment. While increased cerebrospinal fluid (CSF) levels of complement proteins have been observed in patients with AD dementia, no studies have yet investigated the role of complement in the pre‐symptomatic phase of AD, nor throughout its progression. Method We used Luminex technology to assess complement protein levels in 566 CSF samples obtained over five years from 160 high‐risk asymptomatic subjects with familial history of AD [the PREVENT‐AD cohort; Breitner et al, 2016: JPAD 3,(4) 236]. Baseline complement levels were contrasted with CSF biomarkers of AD pathology (Aβ42, p(181) and total tau, NFL) and markers of synaptic dysfunction (GAP‐43, SYT1, SNAP‐25, ADAM22, ADAM23). Analyses were stratified by sex and ApoE4 carrier status. Currently pending are additional longitudinal analyses probing complement levels versus PET‐quantified cerebral amyloid and tau deposition, MRI volume of AD‐implicated structures, and cognitive ability (RBANS). Result Preliminary analyses revealed significant positive associations between complement proteins and p‐(181)tau (C1q: R² = .238, p < .0001; C3: R² = .041, p = .0099; C3b: R² = .027, p = .0384; Factor H: R² = .128, p < .0001) and total tau levels (C1q: R² = .247, p < .0001; C3: R² = .023, p = .0543; Factor H: R² = .128, p < .0001). Complement relationships with synaptic protein levels were more significant overall in females, which showed the following associations [GAP‐43 (C1q: R² = .389, p < .0001; Factor H: R² = .315, p = .0002); SNAP‐25 (C1q: R² = .367, p < .0001; C3: R² = .106, p = .0459; Factor H: R² = .278, p = .0072); SYT1 (C1q: R² = .316, p = .0002; Factor H: R² = .246, p = .0015)]. Conclusion There is a clear association between complement levels, tau pathology, and synaptic markers in the asymptomatic phase of AD, especially in women. Longitudinal and multimodal investigation is ongoing to further characterize these relationships.

Advancements in deep image synthesis techniques, such as generative adversarial networks (GANs) and diffusion models (DMs), have ushered in an era of generating highly realistic images. While this technological progress has captured significant interest, it has also raised concerns about the high challenge in distinguishing real images from their synthetic counterparts. This paper takes inspiration from the potent convergence capabilities between vision and language, coupled with the zero‐shot nature of vision‐language models (VLMs). We introduce an innovative method called Bi‐LORA that leverages VLMs, combined with low‐rank adaptation (LORA) tuning techniques, to enhance the precision of synthetic image detection for unseen model‐generated images. The pivotal conceptual shift in our methodology revolves around reframing binary classification as an image captioning task, leveraging the distinctive capabilities of cutting‐edge VLM, notably bootstrapping language image pre‐training (BLIP)2. Rigorous and comprehensive experiments are conducted to validate the effectiveness of our proposed approach, particularly in detecting unseen diffusion‐generated images from unknown diffusion‐based generative models during training, showcasing robustness to noise, and demonstrating generalisation capabilities to GANs. The experiments show that Bi‐LORA outperforms state of the art models in cross‐generator tasks because it leverages multi‐modal learning, open‐world visual knowledge, and benefits from robust, high‐level semantic understanding. By combining visual and textual knowledge, it can handle variations in the data distribution (such as those caused by different generators) and maintain strong performance across different domains. Its ability to transfer knowledge, robustly extract features and perform zero‐shot learning also contributes to its generalisation capabilities, making it more adaptable to new generators. The experimental results showcase an impressive average accuracy of 93.41% in synthetic image detection on unseen generation models. The code and models associated with this research can be publicly accessed at https://github.com/Mamadou‐Keita/VLM‐DETECT.

The risk of T cell malignancies after chimeric antigen receptor (CAR) T cell therapy is a concern, although the true incidence remains unclear. Here we analyzed the DESCAR-T registry database, encompassing all pediatric and adult patients with hematologic malignancies who received CAR T cell therapy in France since 1 July 2018. Of the 3,066 patients included (2,536 B cell lymphoma, 162 B cell acute lymphoblastic leukemia (ALL) and 368 multiple myeloma), 1,680 (54.8%) received axicabtagene ciloleucel, 205 (6.7%) brexucabtagene autoleucel, 44 (1.4%) lisocabtagene maraleucel and 769 (25.1%) tisagenlecleucel. All multiple myeloma patients received idecabtagene vicleucel, with none receiving ciltacabtagene autoleucel. After a median follow-up of 12.7 months for B cell lymphoma, 17.7 months for B cell ALL and 6.3 months for multiple myeloma, only one (0.03%) patient developed a T cell malignancy after CAR T infusion. Specifically, the patient was diagnosed with a primary cutaneous CD30⁺ T cell lymphoproliferative disorder (anaplastic lymphoma kinase-negative) 3 years after receiving tisagenlecleucel therapy for diffuse large B cell lymphoma. This was associated with the integration of a CAR clone into the tumor suppressor gene PLAAT4 (phospholipase A and acyltransferase 4). Thus, the development of this secondary T cell malignancy might be linked to the use of CAR T cell therapy. In conclusion, our findings indicate a very low risk of T cell malignancy after CAR T cell therapy.