Background: Bloodstream infections continue to pose a serious global health threat due to their high morbidity and mortality, further worsened by rising antimicrobial resistance and delays in starting targeted therapy. This study assesses the accuracy and timeliness of therapeutic recommendations produced by an artificial intelligence (AI)-driven and machine-learning (ML) clinical decision support system (CDSS), comparing results based on molecular diagnostics alone with those that combine molecular and phenotypic data (standard cultures). Methods: In a prospective cross-sectional study conducted in Lima, Peru, 117 blood cultures were analyzed using FilmArray/GeneXpert for molecular identification and MALDI-TOF/VITEK 2.0 for phenotypic profiling. The AI/ML-based CDSS provided treatment recommendations in two formats, which were assessed for concordance and turnaround time. Results: Therapeutic recommendations showed 80.3% consistency between data types, with 86.3% concordance in pathogen and resistance detection. Notably, molecular-only recommendations were delivered 29 h earlier than those incorporating phenotypic data. Escherichia coli was the most frequently isolated pathogen, with a 95% concordance in suggested therapy. A substantial agreement was observed in treatment consistency (Kappa = 0.80). Conclusions: These findings highlight the potential of using AI-powered CDSS in conjunction with molecular diagnostics to accelerate clinical decision-making in bacteremia, supporting more timely interventions and improved antimicrobial stewardship. Further research is warranted to assess scalability and impact across diverse clinical settings.

Background: Antimicrobial stewardship programs (ASPs) are essential in combating antimicrobial resistance (AMR); however, limited resources hinder their implementation. Arkstone, a biotechnology company, developed a machine learning (ML)-driven clinical decision support system (CDSS) to guide antimicrobial prescribing. While AI (artificial intelligence) applications are increasingly used, each model must be carefully validated. Methods: Three components of the ML system were assessed: (1) A prospective observational study tested its ability to distinguish trained from novel data using various validation techniques and BioFire molecular panel inputs. (2) An anonymous retrospective analysis of internal infectious disease lab results evaluated the recognition of novel versus trained complex datasets. (3) A prospective observational validation study reviewed clinical recommendations against standard guidelines by independent clinicians. Results: The system achieved 100% accuracy (F1 = 1) in identifying 111 unique novel data points across 1110 tests over nine training sessions. It correctly identified all 519 fully trained and 644 novel complex datasets. Among 644 clinician-trained reports, there were no major discrepancies in recommendations, with only 100 showing minor differences. Conclusions: This novel ML system demonstrated high accuracy in distinguishing trained from novel data and produced recommendations consistent with clinical guidelines. These results support its value in strengthening CDSS and ASP efforts.

Bacteremia is a life-threatening condition contributing significantly to sepsis-related mortality worldwide. With delayed appropriate antibiotic therapy, mortality increases by 20% regardless of antimicrobial resistance. This study evaluated the perceived clinical utility of Artificial Intelligence (AI)-powered Clinical Decision Support Systems (CDSSs) (OneChoice and OneChoice Fusion) among specialist physicians managing bacteremia cases. A cross-sectional survey was conducted with 65 unique specialist physicians from multiple medical specialties who were presented with clinical vignettes describing patients with bacteremia and 90 corresponding AI-CDSS recommendations. Participants assessed the perceived helpfulness of AI decision-making, the impact of AI recommendations on their own clinical judgment, and the concordance between AI recommendations and their own clinical judgment, as well as the validity of changing therapy based on CDSS recommendations. The study encompassed a diverse range of bacterial pathogens, with Escherichia coli representing 38.7% of the isolates and 30% being extended-spectrum β-lactamase (ESBL) producers. Findings show that 97.8% [(95% CI: 92.2–99.7%)] of physicians reported that AI facilitated decision-making and substantial concordance (87.8% [95% CI: 79.2–93.7%; Cohen’s κ = 0.76]) between AI recommendations and physicians’ therapeutic recommendations. Stratification by pathogen revealed the highest concordance for Escherichia coli bacteremia (96.6%, 28/29 cases). Implementation analysis revealed a meaningful clinical impact, with 68.9% [(95% CI: 58.3–78.2%)] of cases resulting in AI-guided treatment modifications. These findings indicate that AI-powered CDSSs effectively bridge critical gaps in infectious disease expertise and antimicrobial stewardship, providing clinicians with evidence-based therapeutic recommendations that can be integrated into routine practice to optimize antibiotic selection, particularly in settings with limited access to infectious disease specialists. For optimal clinical integration, we recommend that clinicians utilize AI-CDSS recommendations as an adjunct to clinical judgment rather than a replacement, particularly in complex cases involving immunocompromised hosts or polymicrobial infections. Future research should prioritize prospective clinical trials that evaluate direct patient outcomes to establish evidence of broader clinical effectiveness and applicability across diverse healthcare settings.

Background: Persistently, antimicrobial resistance is one of the major global public health challenges, associated with 4.95 million deaths in 2019, of which 1.27 million were directly attributable to resistant bacterial infections. Among all types, urinary tract infections (UTIs) are among the most common worldwide, with 4.49 billion cases in 2021, and they affect older adults the most. In this context, previous studies suggest an exacerbation in the prevalence of resistant pathogens (such as extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae), a circumstance associated with delayed initiation of appropriate treatment, increased mortality, prolonged hospital stays, and selective pressure. These circumstances are more difficult to confront if the limited access to infectious disease specialists is considered, particularly in low- and middle-income countries. In this situation, artificial intelligence (AI) tools have emerged as a potential strategy to support antimicrobial decision-making.Methods: This observational cross-sectional concordance study evaluated the antimicrobial recommendations generated by three AI models: a human-in-the-loop machine learning (ML-HTL) for clinical decision support and two general-purpose language models (GPT-4.0 and Gemini 3.0). A total of 88 cases with positive urine cultures and complete clinical data were included. The AI-generated recommendations for each case were contrasted with the consensus response of an independent panel of three infectious disease specialists. Agreement was assessed using overall concordance, sensitivity, and Cohen’s kappa (κ), with 95% confidence intervals (CI).Results: The mean age of patients was 65.3 years; 81.8% were women, and 80.7% had uncomplicated UTIs. Escherichia coli was the most frequent pathogen (76.1%). Among isolates, 37.5% were ESBL-producing, 53.4% were fluoroquinolone-resistant, and 44.3% were multidrug-resistant. The ML-HTL system showed the highest agreement with the expert´s consensus (90.9%; 95% CI: 83.1–95.3), followed by GPT-4 (63.6%; 95% CI: 53.2–72.9) and Gemini (37.5%; 95% CI: 28.1–47.9). Also, ML-HTL showed high sensitivity and strong agreement (87.5%, κ = 0.89), whereas GPT-4 showed low agreement (κ = 0.32), and Gemini showed less than chance agreement (κ = −0.18).Conclusions: The ML-HTL system, specifically designed for antimicrobial optimization, may better align with experts' consensus than general-purpose language models, supporting their role in clinical decision-making.

Background: Bacteremia is a life-threatening condition contributing significantly to sepsis-related mortality worldwide. With delayed appropriate antibiotic therapy, mortality increases by 20% regardless of antimicrobial resistance. This study evaluated the perceived clinical utility of Artificial Intelligence (AI)-powered Clinical Decision Support Systems (CDSS) (OneChoice and OneChoice Fusion) among specialist physicians managing bacteremia cases. Methods: A cross-sectional survey was conducted with 65 physicians from multiple medical specialties who were presented with clinical vignettes describing patients with bacteremia and corresponding 90 AI-CDSS recommendations. Participants assessed perceived helpfulness, the impact of AI decision-making, concordance between AI recommendations and their own clinical judgment, and whether changing therapy based on CDSS recommendations is valid. Results: The study encompassed a diverse range of bacterial pathogens, with Escherichia coli representing 38.7% of the isolates and 30% being extended-spectrum β-lactamase (ESBL) producers. Results demonstrated 97.8% of physicians reported that AI facilitated decision-making and substantial concordance (87.8%) between AI recommendations and physicians’ therapeutic recommendations. Implementation analysis revealed a meaningful clinical impact, with 68.9% of cases resulting in AI-guided treatment modifications. Conclusions: These findings indicate that AI-powered CDSS effectively bridge critical gaps in infectious disease expertise and antimicrobial stewardship. Future research should prioritize prospective clinical trials evaluating direct patient outcomes to establish evidence of broader clinical effectiveness and application across diverse healthcare settings.

Background: Bloodstream infections remain a critical global health concern due to their high morbidity and mortality, compounded by increasing antimicrobial resistance and delays in initiating targeted therapy. This study evaluates the effectiveness and timeliness of therapeutic recommendations generated by an artificial intelligence (AI)-driven clinical decision support system (CDSS), comparing outputs based solely on molecular diagnostics versus those integrating molecular and phenotypic data. Methods: In a prospective cross-sectional study conducted in Lima, Peru, 117 blood cultures were analyzed using FilmArray/GeneXpert for molecular identification and MALDI-TOF/VITEK 2.0 for phenotypic profiling. The AI-based CDSS provided treatment recommendations in two formats, which were assessed for concordance and turnaround time. Results: Therapeutic recommendations showed 80.3% consistency between data types, with 86.3% concordance in pathogen and resistance detection. Notably, molecular-only recommendations were delivered 29 hours earlier than those incorporating phenotypic data. Escherichia coli was the most frequently isolated pathogen, with a 95% concordance in suggested therapy. A substantial agreement was observed in treatment consistency (Kappa = 0.80). Conclusions: These findings highlight the potential of AI-powered molecular diagnostics to accelerate clinical decision-making in bacteremia, supporting more timely interventions and improved antimicrobial stewardship. Further research is warranted to assess scalability and impact across diverse clinical settings.