Biosensors Conference

Abstract

Background and aim: The suppression of the ovaries could be related to COVID-19, as they are often sup- pressed in acute illnesses to ensure the normal functioning of other systems and organs. In this context, some patients with COVID-19 were found to have inappropriately high levels of follicle-stimulating hormone (FSH) and luteinizing hormone (LH) [1,2]. In the immune response against SARS-CoV-2, interleukin-6 is released along with other cytokines such as interleukin-8 and tumor necrosis factor alpha, which trigger a pro-coagulant state that is unfavorable for the blastocyst or fetus in a normal human uterus. In addition, high rates of antiphospho- lipid antibodies (aPLs) have been found in severe COVID-19 patients, although their association with intrahos- pital vascular thrombosis or worse outcome was not clearly proven [3], and also in patients with subsequent post-COVID syndrome, which could affect pregnancy and fertility. For diagnosis of antiphospholipid syndrome (APS), vascular thrombosis should be accompanied with the presence of aPLs. The aim of this study was to investigate the presence of aPLs in severe COVID-19 female patients and to establish their association with the actual vascular thrombosis, thrombosis in personal history and pregnancy complications.

Methods: In 35 female patients hospitalized with severe COVID-19 pneumonia from March to May 2022, vas- cular thrombosis and pregnancy complications were recorded during the entire hospital stay and in their med- ical history. Females were also tested for nine types of aPLs at 4 time points (admission, exacerbation and discharge from hospital and at 3-month follow-up): anticardiolipin (aCL), anti-β2-glycoproteinI (anti-β2GPI) and antiphosphatidylserine/prothrombin (aPS/PT) of the IgM/IgG/IgA isotypes. Because we used anticoagulants at therapeutic doses, we could not use the LA test, even with neutralizers, as these are only effective at low and typically preventive LMWH doses (0.8–1.0 U/mL); instead, the aPS/PT test was performed. Positive values were defined as those above the 99th percentile of the healthy control population. Statistical analysis was per- formed using SPSS 21 (NY, USA IBM Corp. Released 2021, IBM SPSS Statistics for Windows).

Results: During hospitalization, aPLs were tested positive at least once in 51.4% of females. All 5.7% of patients that passed away were tested aPLs negative at admission and worsening (one had pulmonary artery throm- bosis and one microthrombosis). Personal anamnesis (PA) for thromboembolism was verified for 4 patients, at admission all of them tested aPLs negative, however transition to aPLs positivity at discharge (as the dis- ease subsided) was detected in 75% (all with deep vein thrombosis in PA) due to newly- positive aCL IgG at discharge seen in 75% added with positive aCL IgM in 50%. At follow-up, one female remained positive for aCL IgG, one negativized and one one did not appear. None of the women in our cohort were pregnant, and none of them became pregnant while at the hospital and in the follow-up. In their personal history, two women had abortions after 10th week of gestation, but none of them tested positive for aPLs at admission or follow-up.

Conclusion: The incidence of APS in our cohort was 2.8%, whereas in general population it ranges from 0.001% to 0.002%. This high incidence was mostly due to deep vein thrombosis in womens personal history. Therefore, it is advised that all females with personal anamnesis of deep vein thrombosis in the past who had suffered severe COVID-19 and who plan to conceive should be tested for aPLs once and in positive cases retested 3 months later, in order to rule out the APS diagnosis and order appropriate therapy during pregnancy. Also, sup- pression of the ovaries related to COVID-19 in the post-COVID syndrome could be the reason for infertility [4].

Abstract

Tumor-derived exosomes (TDEs) are key mediators of intercellular communication in cancer, transferring on- cogenic cargo: proteins, nucleic acids, and metabolites—to recipient cells within local and distant microenvi- ronments. By disseminating carcinogenic materials, TDEs drive immunosuppression, inflammation, angiogen- esis, treatment resistance, epithelial- mesenchymal transition (EMT), and metastasis. Critically, TDEs can also induce malignant transformation in otherwise normal cells, highlighting their central role in cancer progression. Consequently, eliminating TDEs represents a promising strategy to augment current cancer therapies and curb metastatic spread. However, existing approaches to block TDE release or uptake remain ineffective and often incur off-target toxicities. Given the structural and biochemical similarities between TDEs and viruses—includ- ing size, membrane composition, and biogenesis—we propose adapting virus-engineered methodolo- gies to develop vaccines targeting TDEs (anti-TDE vaccines). This strategy involves genetically engineered tumor cells to generate modified, immunogenic TDEs that are incapable of binding to or entering recipient cells. Specifical- ly, we advocate for: Surface Engineering: Decorating TDE membranes with foreign antigens to provoke potent, durable immune responses against TDEs, enabling their systemic clearance. Gene Downregulation: Suppress- ing expression of specific transmembrane proteins (e.g., tetraspanins, integrins) in parent tumor cells to gener- ate TDEs that lack functional ligands required for target cell attachment and internalization.

Further, we refer TDEs as local viruses (loruses) to reflect their endogenous origin and their involvement in the trafficking of pathological information. Engineered loruses could serve as effective vaccines, priming the im- mune system to recognize and eliminate natural TDEs while avoiding unintended toxicity. This approach merg- es insights from onco-immunology and virology to overcome the challenges of TDE-mediated pathogenesis and open new avenues for cancer immunotherapy.

Abstract

The accelerating convergence of Artificial Intelligence (AI), Blockchain, and Zero Trust Architecture (ZTA) is redefining the foundations of modern cybersecurity. As digital ecosystems become increasingly decentral- ized, dynamic, and heterogeneous spanning IoT, edge, and cloud infrastructures traditional perimeter-based defences are proving insufficient against emerging smart attacks that leverage automation, data poisoning, and adversarial learning. This presentation explores an integrated AI-Blockchain paradigm for implementing adaptive Zero Trust frameworks capable of self-learning, continuous verification, and autonomous response. The session begins with a critical overview of current research trends in AI-driven ZTA and blockchain-enabled trust management, highlighting persistent challenges related to scalability, latency, and energy efficiency in constrained environments. It then introduces a unified conceptual model in which AI delivers behavioural ana- lytics, dynamic trust scoring, and predictive threat intelligence, while blockchain ensures decentralised identity, immutable auditability, and policy enforcement through smart contracts. The synergy of these technologies enables a cognitive trust fabric a distributed cyber-defence architecture that continuously authenticates enti- ties and validates actions in real time. Drawing upon recent empirical studies and the speaker’s own research contributions including the IbiboRPLChain solution for blockchain-based authentication in IoT-LLNs the presen- tation demonstrates measurable improvements in security, consensus stability, and computational efficiency. The talk further discusses practical use-cases across critical infrastructures, industrial IoT, and smart cities, and outlines a roadmap for integrating Proof-of-Authority consensus, federated AI, and post-quantum cryptog- raphy into future Zero Trust ecosystems. Ultimately, this work argues that the fusion of AI’s cognitive adapt- ability with blockchain’s cryptographic assurance represents a pivotal step toward autonomous, verifiable, and resilient cyber defences. In the era of intelligent adversaries, trust must not be assumed it must be continuously earned, and cryptographically proven.

Abstract

Depression and anxiety often co-occur yet differ markedly in their behavioral and vocal expression. While de- pression is typically characterized by slowed, flattened speech and reduced prosody, anxiety manifests more variably — ranging from tense and rapid to inhibited vocal patterns — making it harder to identify reliably in clin- ical encounters. In this study, we analyzed voice samples from 60 outpatients (30 high, 30 low per condition) whose depression and anxiety levels were independently assessed via clinician (MD) evaluation, self-report questionnaires (CES-D-6, GAD-7), and self-reported SUDS scales (subjective units of depression and anxiety). A three-fold cross-validation of a voice-based classifier revealed that SUDS ratings showed the strongest cor- respondence with acoustic features (AUC ≈ 0.69 for anxiety), followed by MD-rated depression (AUC ≈ 0.61). In contrast, self-report questionnaires demonstrated weak alignment (AUC ≈ 0.30 – 0.56). Notably, clinicians’ anx- iety ratings correlated less strongly with vocal markers than their depression ratings, suggesting that medical doctors tend to be less adept at identifying anxiety among their patients, often equating it with overt agitation or “flustered” behavior. Our findings indicate that voice analytics may capture latent depression and anxiety signals overlooked in traditional evaluations, underscoring the potential of an objective, data-driven triage tool to complement clinical judgment in mental-health screening.