Deep learning application to automated classification of recommendations made by hospital pharmacists during medication prescription review.

PURPOSE: Recommendations to improve therapeutics are proposals made by pharmacists during the prescription review process to address suboptimal use of medicines. Recommendations are generated daily as text documents but are rarely reused beyond their primary use to alert prescribers and caregivers. If recommendation data were easier to summarize, they could be used retrospectively to improve safeguards for better prescribing. The objective of this work was to train a deep learning algorithm for automated recommendation classification to valorize the large amount of recommendation data. METHODS: The study was conducted in a French university hospital, at which recommendation data were collected throughout 2017. Data from the first 6 months of 2017 were labeled by 2 pharmacists who assigned recommendations to 1 of the 29 possible classes of the French Society of Clinical Pharmacy classification. A deep neural network classifier was trained to predict the class of recommendations. RESULTS: In total, 27,699 labeled recommendations from the first half of 2017 were used to train and evaluate a classifier. The prediction accuracy calculated on a validation dataset was 78.0%. We also predicted classes for unlabeled recommendations collected during the second half of 2017. Of the 4,460 predictions reviewed, 67 required correction. When these additional labeled data were concatenated with the original dataset and the neural network was retrained, accuracy reached 81.0%. CONCLUSION: To facilitate analysis of recommendations, we have implemented an automated classification system using deep learning that achieves respectable performance. This tool can help to retrospectively highlight the clinical significance of daily medication reviews performed by hospital clinical pharmacists.

Using machine learning or deep learning models in a hospital setting to detect inappropriate prescriptions: a systematic review.

OBJECTIVES: The emergence of artificial intelligence (AI) is catching the interest of hospital pharmacists. A massive collection of health data is now available to train AI models and hold the promise of disrupting codes and practices. The objective of this systematic review was to examine the state of the art of machine learning or deep learning models that detect inappropriate hospital medication orders. METHODS: A systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. MEDLINE and Embase databases were searched from inception to May 2023. Studies were included if they reported and described an AI model intended for use by clinical pharmacists in hospitals. Risk of bias was assessed using the Prediction model Risk Of Bias ASsessment Tool (PROBAST). RESULTS: 13 articles were selected after review: 12 studies were judged to have high risk of bias; 11 studies were published between 2020 and 2023; 8 were conducted in North America and Asia; 6 analysed orders and detected inappropriate prescriptions according to patient profiles and medication orders; and 7 detected specific inappropriate prescriptions, such as detecting antibiotic resistance, dosage abnormality in prescriptions, high alert drugs errors from prescriptions or predicting the risk of adverse drug events. Various AI models were used, mainly supervised learning techniques. The training datasets used were very heterogeneous; the length of study varied from 2 weeks to 7 years and the number of prescription orders analysed went from 31 to 5 804 192. CONCLUSIONS: This systematic review points out that, to date, few original research studies report AI tools based on machine or deep learning in the field of hospital clinical pharmacy. However, these original articles, while preliminary, highlighted the potential value of integrating AI into clinical hospital pharmacy practice.

Evidence of dose-response following hypofractionated stereotactic radiotherapy to the cavity after surgery for brain metastases.

BACKGROUND AND OBJECTIVE: A retrospective review of consecutive patients between January 2012 and December 2018 receiving hypofractionated stereotactic radiotherapy (HSRT) to the cavity after resection for brain metastases was performed. METHODS: Treatment was delivered using an appropriately commissioned linear accelerator. The primary outcome was time to radiological or histological confirmation of local recurrence following completion of HSRT. Dose-fractionation regimens were converted to biologically 2 Gy-equivalent doses assuming α/ß = 10 (EQD2[10]). Multivariate Cox proportional hazards modelling was performed to determine hazard ratios (HR) with respective 95% confidence intervals (CI). The Log-rank test was used to determine p values taking statistical significance p < 0.05. RESULTS: There were 134 patients and 144 cavities identified. The most common primary histologies were melanoma (n = 49) and lung (n = 32). 116 patients (87%) underwent a gross total resection. Median planning target volume (PTV) was 28 cm3 (range 2.4-149.2). Median EQD2[10] was 38.4 Gy (range 22.3-59.7) and 24 Gy in 3 fractions was the most common regimen. 12 (9%) patients demonstrated local recurrence at median interval 215 days (range 4-594). 7 (5%) patients experienced grade 3 or higher toxicities. In multivariate analysis, EQD2[10] was associated with local failure such that increased equivalent doses improved local control [HR = 0.79 and 95% CI 0.65-0.96, p = 0.0192]. There were no significant associations for primary histology, patient age, volume of residual disease, PTV volume or location. CONCLUSION: This large series demonstrates that HSFRT to the surgical resection cavity for brain metastases has improved local control with increasing dose. Rates of grade 3 or higher toxicity were low overall.

Early diagnosis of mild cognitive impairment and Alzheimer's with event-related potentials and event-related desynchronization in N-back working memory tasks.

BACKGROUND AND OBJECTIVE: In this study we investigate whether or not event-related potentials (ERP) and/or event-related (de)synchronization (ERD/ERS) can be used to differentiate between 27 healthy elderly (HE), 21 subjects diagnosed with mild cognitive impairment (MCI) and 15 mild Alzheimer's disease (AD) patients. METHODS: Using 32-channel EEG recordings, we measured ERP responses to a three-level (N-back, N = 0,1,2) visual working memory task. We also performed ERD analysis over the same EEG data, dividing the full-band signal into the well-known delta, theta, alpha, beta and gamma bands. Both ERP and ERD analyses were followed by cluster analysis with correction for multicomparisons whenever significant differences were found between groups. RESULTS: Regarding ERP (full-band analysis), our findings have shown both patient groups (MCI and AD) with reduced P450 amplitude (compared to HE controls) in the execution of the non-match 1-back task at many scalp electrodes, chiefly at parietal and centro-parietal areas. However, no significant differences were found between MCI and AD in ERP analysis whatever was the task. As for sub-band analyses, ERD/ERS measures revealed that HE subjects elicited consistently greater alpha ERD responses than MCI and AD patients during the 1-back task in the match condition, with all differences located at frontal, central and occipital regions. Moreover, in the non-match condition, it was possible to distinguish between MCI and AD patients when they were performing the 0-back task, with MCI presenting more desynchronization than AD on the theta band at temporal and fronto-temporal areas. In summary, ERD analyses have revealed themselves more valuable than ERP, since they showed significant differences in all three group comparisons: HE vs. MCI, HE vs. AD, and MCI vs. AD. CONCLUSIONS: Based on these findings, we conclude that ERD responses to working memory (N-back) tasks could be useful not only for early MCI diagnosis or for improved AD diagnosis, but probably also for assessing the likelihood of MCI progression to AD, after further validated by a longitudinal study.

The profile of executive functioning in amnestic mild cognitive impairment: disproportionate deficits in inhibitory control.

Amnestic mild cognitive impairment (aMCI) represents a group of individuals who are highly likely to develop Alzheimer's disease (AD). Although aMCI is typically conceptualized as involving predominantly deficits in episodic memory, recent studies have demonstrated that deficits in executive functioning may also be present, and thorough categorization of cognitive functioning in MCI may improve early diagnosis and treatment of AD. We first provide an extensive review of neuropsychology studies that examined executive functioning in MCI. We then present data on executive functioning across multiple sub-domains (divided attention, working memory, inhibitory control, verbal fluency, and planning) in 40 aMCI patients (single or multiple domain) and 32 normal elderly controls (NECs). MCI patients performed significantly worse than NECs in all 5 sub-domains, and there was impairment (>1.0 SD below the mean of NECs) in all sub-domains. Impairment on each test was frequent, with 100% of MCI patients exhibiting a deficit in at least one sub-domain of executive functioning. Inhibitory control was the most frequently and severely impaired. These results indicate that executive dysfunction in multiple sub-domains is common in aMCI and highlights the importance of a comprehensive neuropsychological evaluation for fully characterizing the nature and extent of cognitive deficits in MCI.

Executive functions in frontotemporal dementia and Lewy body dementia.

Diagnosis of different types of dementia is often based on clinical symptomatology rather than underlying pathology; therefore, accurate diagnosis depends on a thorough description of cognitive functioning in different dementias. Furthermore, direct comparison of cognitive functions between different types of dementia is necessary for differential diagnosis. Executive dysfunction is common in several types of dementia, including frontotemporal dementia (FTD) and Lewy body dementia (LBD); however, FTD and LBD patients have never been directly compared on measures of executive functioning. The authors compared the performance of 17 FTD and 15 LBD patients on 6 measures of executive functioning in terms of statistical group differences, mean severity of clinical impairment in comparison to normal controls, and frequency of impairment. Results indicated a remarkably similar pattern of performance across all areas examined in terms of mean performance, as well as degree and frequency of impairment. Only the Stroop test produced results that could potentially differentiate the patient groups. These findings suggest that both FTD and LBD should be considered disorders involving executive dysfunction.

Clinical characteristics of bipolar disorder in very young children.

BACKGROUND: Clinical information about bipolar disorder (BPD) in preschool-age (3-7 years old) children is extremely limited. This study examined clinical presentations, applicability of the DSM-IV diagnostic criteria, comorbidity, recovery and relapse rates, as well as some treatment strategies used in the management of BPD in preschoolers. METHODS: The charts of 26 outpatient children, ages 3-7, refereed to a child psychiatry outpatient clinic with mood and behavioral symptoms, were retrospectively reviewed. RESULTS: The majority of the patients were referred with the tentative diagnosis of ADHD but the most common diagnoses made by child and adolescent psychiatrists at the time of initial evaluation were BPD NOS (61.5%), followed by BPD I (26.9%), and mood disorder NOS (23.1%). Thirty-eight percent of the patients had one or more comorbid diagnoses. The most common presenting symptoms were irritability (84.6%) and aggression (88.5%). The most widely prescribed class of medications after diagnosis in the clinic was atypical antipsychotics and mood stabilizers. Twenty-six percent of the patients were treated with a combination of atypical antipsychotics and mood stabilizers. LIMITATIONS: Retrospective design; small sample size; lack of a comparison group. CONCLUSIONS: The course of BPD with onset in preschool years is complicated with high recovery and relapse rates. The questions of development of age-appropriate diagnostic criteria, long-term prognosis and treatment strategies used in this population require further intensive investigation.

Adjunctive quetiapine for treatment-resistant adolescent major depressive disorder: a case series.

BACKGROUND: Major depressive disorder (MDD) is a leading cause of disability and mortality in adolescents. Empirical evidence suggests that many adolescents with MDD do not respond or respond only partially to commonly used interventions (antidepressants, psychotherapy, or a combination of antidepressants and psychotherapy). There is preliminary data in adults that adjunctive second-generation antipsychotics may be useful in treatment-resistant depression. OBJECTIVE: The aim of this study was to obtain preliminary data regarding the safety, tolerability, and clinical usefulness of quetiapine as adjunctive therapy for adolescents (13-18 years of age) diagnosed with treatment-resistant MDD. Treatment-resistant MDD was defined as a failure to respond to an adequate dose for at least 8 weeks of a selective serotonin reuptake inhibitor (SSRI). METHODS: The medical charts of 10 adolescents (13-18 years of age) diagnosed with treatment- resistant MDD, who were treated with adjunctive quetiapine, were evaluated. Doses of preexisting antidepressants remained unchanged during the period of evaluation. Response to treatment was defined as a Clinical Global Impression-Improvement (CGI-I) score of 1 (very much improved) or 2 (much improved). RESULTS: Seven adolescents (70%) qualified as responders to treatment with adjunctive quetiapine. The median dose of quetiapine was 200 mg (mean +/- SD = 275 +/- 190.4 mg, range; 150-800 mg). Side effects included sedation (40%) and weight gain (mean +/- SD = 4.5 +/- 7.24 pounds). There was no serious adverse event. CONCLUSIONS: This case series suggests that there may be a role for adjunctive quetiapine in treatment-resistant adolescent depression. Clinical safety and efficacy trials of quetiapine in this population appear to be warranted.