National Cancer Center of Georgia

This work provides an analytical and theoretical–practical justification for the effectiveness of deploying adaptive retraining mechanisms for machine learning models in unstable industrial environments that evolve dynamically over time. The relevance of the study is driven both by the sharp increase in investment in artificial intelligence in 2025 and by the objective need to curb losses in model accuracy caused by concept drift and input data drift. The study adopts a comprehensive methodological approach that combines a systematic review of scientific publications, a comparative analysis of drift-detection algorithms, and a case study component. The results indicate that introducing automated process orchestration using Apache Airflow, together with monitoring systems that provide full model coverage, reduces incident detection time by 60% and lowers the labour intensity of data preparation for subsequent retraining by 40%. An original hybrid lifecycle governance model for ML systems is proposed, combining statistical activation mechanisms with business-oriented principles to prioritise computational resources. The conclusions empirically support the hypothesis that a meaningful increase in model accuracy—up to 11% in marketing attribution tasks—can be achieved while simultaneously reducing operating costs. The study’s materials carry both practical and scientific value for machine learning practitioners, data architects, and digital unit leaders responsible for scaling and ensuring the resilient operation of AI solutions in production environments.

Negative emotions and thoughts and physical changes such as an increased level of blood pressure characterize anxiety. Depressive disorder is when a person loses interest or has little interest in doing day-to-day activities, and it can affect all parts of life. The person can also experience mood swings and suicidal and thoughts of self-harm. The purpose of this study is to understand anxiety and depression in medical students and the correlation of issues such as academic pressure, bullying, and the hostile atmosphere in medical schools along with the resources provided by the medical schools and if they are adequate to deal with anxiety and depression among the students. The study also reviews suicidal and self-harm tendencies among medical students.

BackgroundOlutasidenib is an oral, selective inhibitor of mutant isocitrate dehydrogenase 1 (mIDH1), FDA-approved for relapsed/refractory (R/R) acute myeloid leukemia (AML) based on a registrational, phase 2, open-label, multicenter trial.MethodsResults from the pre-planned interim analysis were previously published (data cut-off [DCO]: June 2021). In this final-follow up analysis, we report an additional 2 years of efficacy and safety data (DCO: June 2023).ResultsAt study completion, the overall population included 153 patients (median age, 71 years); 66% had received ≥ 2 prior treatment regimens, and 39% with a hypomethylating agent. Among the 147 efficacy-evaluable patients, 51 achieved complete remission (CR) or CR with partial hematologic recovery (CRh), resulting in a CR/CRh rate of 35% (P < 0.001; 95% CI, 27–43), with 32% of responders achieving CR. The median time to CR/CRh was 1.9 months (range, 0.9–5.6 months). Among responders, 33% achieved CR/CRh within 2–4 months and 12% required ≥ 4 months. The overall response rate (ORR) was 48% (n = 71; 95% CI, 40–56.7). Median duration of CR/CRh was 25.3 months (95% CI, 13.5–not reached), and median overall survival (OS) was 11.5 months (95% CI, 8.3–15.5). Patients with 1–2 prior regimens had a higher CR/CRh rate (41%) and longer median OS (13 months) than those with ≥ 3 prior regimens (CR/CRh: 24%; median OS: 8.9 months). CR/CRh rates were higher among patients with R132C (42%) and R132L/G/S mutations (33%) compared with those harboring R132H mutations (17%). Response rates decreased with increasing numbers of co-mutations. Few new adverse events (AEs) and no treatment discontinuations due to AEs occurred beyond Year 3.ConclusionThese 5-year data support the durable efficacy and manageable safety profile of olutasidenib in R/R mIDH1 AML, including heavily pretreated patients. Findings highlight the potential role of olutasidenib in earlier lines of treatment, and support sustaining therapy for at least 6 months to allow for a clinical response. Further research is warranted to optimize treatment sequencing and patient selection.Trial registrationNCT02719574.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13045-025-01751-w.

BACKGROUNDSimilar to many low‐ and middle‐income countries, the barriers limiting wider mechanical thrombectomy (MT) access in Georgia are largely unknown. Recently, the MT access score (MTAS) was introduced as a new tool for identifying and characterizing barriers to MT access. This study aimed to implement the MTAS in Georgia, a middle‐income country in Eastern Europe, to assess and characterize national barriers to MT.METHODSWe applied the MTAS, which comprises 12 weighted attributes, each scored on a 0–3 scale, resulting in a total score range of 0–36, 0 being the worst possible score. Eight members of the Mission Thrombectomy regional committee from different regions of Georgia were invited as panelists in this survey. The results of the survey are shown as a median with an interquartile range.RESULTSThe median MTAS for Georgia was 17. The lowest median scores were documented for 2 attributes: lack of prehospital large vessel occlusion‐specific screening [0.0 (0.0–0.0)] and telestroke networks [0.0 (0.0–0.0)], with 87.5% of panelists assessing the score as 0 for both attributes. The highest scores were obtained for emergency medical services use [3.0 (2.0–3.0)], availability of MT operators [2.0 (2.0–2.5)] followed by MT device availability and government/insurance coverage [2.0 (2.0–2.0) for each].CONCLUSIONMTAS is a valid tool for quantitatively assessing barriers to MT in Georgia. It identified a lack of information and the presence of physical barriers as major challenges. These findings underscore the need for targeted interventions through national stroke public health initiatives to improve access to MT.

6516 Background: ASC2ESCALATE (NCT05384587) is the first prospective trial of ASC in CML-CP after 1 TKI (2L) with dose escalation for pts with suboptimal response; ASC is also being assessed in a separate newly diagnosed (1L) cohort. A previous IA of the 2L cohort reported ASC’s safety (n=71) and wk 24 efficacy (n=28; BCR::ABL1 IS ≤1%, 85.7%; major molecular response [MMR], 42.9%). We report updated safety (n=101) and wk 24 efficacy (n=63) results. Methods: ASC2ESCALATE is a phase 2, single-arm, open-label US study of ASC in adults with 1L or 2L CML-CP without the T315I mutation. In the 2L cohort, eligible pts had discontinued their prior TKI due to warning or failure per ELN2020 or intolerance with BCR::ABL1 IS &gt;0.1% at screening. Pts received ASC 80 mg once daily (QD). If BCR::ABL1 IS &gt;1% at wk 24, dose was increased to 200 mg QD. If BCR::ABL1 IS &gt;0.1% at wk 48, dose was increased from 80 to 200 mg QD or from 200 mg QD to 200 mg twice daily, or pts could be taken off study. If pts had any grade 3/4 or persistent grade 2 toxicity refractory to optimal management, they were ineligible for dose escalation at wk 24 and/or 48 and continued the same dose. Results: This IA included all 101 pts enrolled with 2L CML-CP; all pts had received ≥1 ASC dose by the cutoff (Nov 15, 2024). Prior treatment (Tx) included dasatinib (44.6%), imatinib (42.6%), nilotinib (9.9%), or bosutinib (5.0%); 66.3% of pts had received prior Tx for ≥12 mo. Pts discontinued prior Tx due to lack of efficacy (56.4%) or intolerance (43.6%). By the cutoff, 92 pts (91.1%) remained on ASC; 9 pts (8.9%) discontinued ASC, mostly due to adverse events (AEs; n=4) and pt decision (n=3). Median duration of ASC exposure was 26.1 (range, 6-100) wk. Pts evaluable for all efficacy analyses completed assessments for the respective timepoint or discontinued earlier (wk 4, n=94; wk 12, n=86; wk 24, n= 63). At wk 4, 12, and 24, 46.8%, 84.9%, and 82.5% pts, respectively, had BCR::ABL1 IS ≤1%. Deeper responses were also achieved at wk 12 (MMR, 39.5%; MR 4 , 11.6%; MR 4.5 , 2.3%) and 24 (MMR, 44.4%; MR 4 , 25.4%; MR 4.5 , 9.5%). Seven pts had dose escalation from 80 to 200 mg QD per their response level at wk 24 (n=3) and 48 (n=4). All-grade AEs ≥20% were headache (22.8%) and nausea (20.8%). Grade ≥3 AEs ≥5% were hypertension (8.9 %), thrombocytopenia (6.9%), and neutropenia (5.9%). AEs led to dose adjustment/interruption in 27 pts (26.7%). AEs led to discontinuation in 4 pts; 1 of these AEs occurred &gt;30 d after last ASC dose. No arterial-occlusive events or on-Tx deaths occurred. Conclusions: 2L ASC demonstrated high molecular response rates at wk 24 and safety consistent with previously established ASC data across Tx lines; no new or worsening safety signals arose. ASC was tolerable with few AEs leading to discontinuation. These IA results support ASC as a Tx option in 2L CML-CP. The impact of dose escalation continues to be explored. Clinical trial information: NCT05384587 .

BackgroundOlutasidenib is a potent, selective, oral, small molecule inhibitor of mutant IDH1 (mIDH1) which induced durable remissions in high-risk, relapsed/refractory (R/R) mIDH1 AML patients in a phase 1/2 trial. We present a pooled analysis from multiple cohorts of the phase 1/2 trial of patients with R/R AML who received combination olutasidenib and azacitidine therapy.MethodsAdult patients with mIDH1R132 AML received 150 mg olutasidenib twice daily plus standard-of-care azacitidine (OLU + AZA) and were evaluated for response and safety.ResultsSixty-seven patients with R/R mIDH1R132 AML received combination OLU + AZA. Median age was 66 years (range 28–82) and 54% were male. Most patients (83%) had 2 + prior regimens, including a hypomethylating agent in 40%, IDH1 inhibitor therapy in 31% (olutasidenib in 24%), and hematopoietic stem cell transplant in 10%. Cytogenetic risk was intermediate in 72%, poor in 18% and unknown in 10%. CR/CRh was achieved in 21/67 (31%; 95% CI 21–44) patients, with a median duration of 14.7 months (95% CI 4.6-not reached). CR was achieved in 18/67 (27%; 95% CI 17–39) patients, with median duration of 20.3 months (95% CI 3.7-not reached). Overall response (partial remission or better) was achieved in 34/67 (51%; 95% CI 38–63) patients. Median overall survival was 12.9 months (95% CI 18.7–19.3). In a subset analysis excluding patients who had prior OLU exposure (N = 51), CR/CRh was achieved in 19/51 (37%; 95% CI 24–52) patients, CR was achieved in 16/51 (31%; 95% CI 19–46), and overall response was achieved in 30/51 (59%; 95% CI 44–72). In patients who achieved CR/CRh and were transfusion-dependent at baseline, transfusion independence (RBC and platelets) was achieved in 64% (7/11) and 57% (4/7) of patients, respectively. The most common Grade 3 or 4 adverse events (> 20% patients) were decreased platelet count (37%), red blood cell count (25%), and neutrophil count (24%). Six patients (9%) experienced differentiation syndrome. Four (6%) discontinued treatment due to an adverse event.ConclusionsOlutasidenib plus azacitidine induced high response rates and durable remissions with a tolerable side effect profile in patients with R/R AML with diverse treatment histories. The results represent another therapeutic option for patients with mIDH1 AML who may benefit from a targeted therapy.Trial registrationNCT02719574.

Bosutinib is a second-generation tyrosine kinase inhibitor (TKI) approved for use in patients with newly diagnosed Philadelphia chromosome (Ph)-positive chronic phase (CP) chronic myeloid leukemia (CML), as well as Ph-positive CP, accelerated phase, or blast phase (with chemotherapy) CML resistant or intolerant to prior therapy. Clinical trials have shown bosutinib is effective as first-line therapy for patients with CML as well as in later lines of therapy after prior TKI failure. Bosutinib has an established safety profile; however, as with all TKIs approved for the treatment of CML, there are adverse events (AEs) that require management. The safety profile of bosutinib is characterized by gastrointestinal, hematological, hepatic, and skin toxicities. Many of these AEs can be managed with dose adjustment strategies. In this podcast, the authors summarize data from some recent bosutinib publications and discuss implications for optimizing bosutinib treatment of patients with CML.Podcast Video (MP4 210846 KB)Supplementary InformationThe online version contains supplementary material available at 10.1007/s11523-024-01123-3.

The objective of this study was to evaluate the effectiveness of human papillomavirus HPV test with HPV16/18 genotyping and liquid-based cytology (LBC) triage as a primary screening method for cervical cancer compared to conventional Pap test in women undergoing routine cervical cancer screening in Tbilisi. Cross-sectional, prospective study was conducted, where 1,000 enrolled women aged 30-60 years during one visit underwent conventional Pap smear and Hr-HPV testing (Roche Cobas system). Women with any positive screening results were referred for further evaluation and remaining cells from the Cell Collection Medium vial were used for LBC. The study calculated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for each screening method and receiver-operating characteristic (ROC) curve to evaluate the accuracy of each diagnostic method in identifying people with CIN2+ diseases. The HPV test with HPV16/18 genotyping and LBC triage demonstrated higher sensitivity (76.9%), specificity (71.6%), and PPV (34.5%) compared to conventional Pap tests (p < 0.05). NPV was also high with the HPV test (94.1%). The HPV test alone had the highest sensitivity (92.3%) and NPV (96.7%), but lower specificity (41.4%) and PPV (22.6%) than the HPV test with HPV16/18 genotyping and LBC triage (p < 0.05). Comparing the areas under the curve (AUCs), only the HPV with HPV16/18 genotyping and LBC triage showed a statistically significant difference when compared to conventional Pap (0.71 vs. 0.55, p = 0.03) and high figures of AUC 0.71 (95% CI: 0.58-0.85) suggesting that HPV test with HPV16/18 genotyping and LBC triage is a more reliable screening method for detecting CIN2+ disease and preventing cervical cancer, than other screening modality. The results suggest that the HPV test with HPV16/18 genotyping and LBC triage is a more effective primary screening method compared to conventional Pap tests. This information should be the basis for transition from cytological screening to HPV testing in Georgia.