Search

Abstract Diagnosing patients in the medical emergency department is complex and this is expected to increase in many countries due to an ageing population. In this study we investigate the feasibility of training machine learning algorithms to assist physicians handling the complex situation in the medical emergency departments. This is expected to reduce diagnostic errors and improve patient logistics and outcome. We included a total of 9,190 consecutive patient admissions diagnosed and treated in two hospitals in this cohort study. Patients had a biochemical workup including blood and urine analyses on clinical decision totaling 260 analyses. After adding nurse-registered data we trained 19 machine learning algorithms on a random 80% sample of the patients and validated the results on the remaining 20%. We trained algorithms for 19 different patient outcomes including the main outcomes death in 7 (Area under the Curve (AUC) 91.4%) and 30 days (AUC 91.3%) and safe-discharge(AUC 87.3%). The various algorithms obtained areas under the Receiver Operating Characteristics -curves in the range of 71.8–96.3% in the holdout cohort (68.3–98.2% in the training cohort). Performing this list of biochemical analyses at admission also reduced the number of subsequent venipunctures within 24 h from patient admittance by 22%. We have shown that it is possible to develop a list of machine-learning algorithms with high AUC for use in medical emergency departments. Moreover, the study showed that it is possible to reduce the number of venipunctures in this cohort.

Abstract This study aimed to assess how parents perceived treatment‐related side effects during acute lymphoblastic leukemia (ALL) treatment. Parents of children 1–17.9 years at diagnosis in Sweden, Finland, and Denmark who were alive and in first remission ≥6 months after end of ALL treatment were asked to respond on specific items regarding how their child was affected by side effects related to vincristine (VCR), corticosteroids, peg‐asparaginase (ASP), and maintenance therapy, as well as overall impact of these treatments, complications in general, and their perception of impact on their child in comparison with other children with ALL. Parents of 307 children responded. More than a third reported that their child had been affected to a high extent by VCR (39.7%) and corticosteroids (35.8%), with walking difficulties, muscular weakness, pain, changes in appetite, and mood swings as the most common and severe symptoms. Reporting of these toxicities was lacking from the NOPHO ALL2008 database, except for peripheral paralysis (12.1%). For distinct toxicities reported in the NOPHO ALL2008 database, for example, thrombosis and pancreatitis, parent reports were similar to the database. Although a high overall negative impact during treatment was reported, parents generally rated the impact on their child as less, or similar, to other children with ALL. Parents perceived VCR and corticosteroid therapy, in particular, to have a negative impact on their child during ALL treatment, which was not captured in the NOPHO ALL2008 toxicity reporting. Our results highlight the importance of including patient/parent‐reported outcomes in toxicity reporting.

Risk-adapted treatment in acute lymphoblastic leukemia (ALL) relies on genetic information and measurable residual disease (MRD) monitoring. In this proof of concept study, DNA from diagnostic bone marrow (BM) of six children with ALL, without stratifying genetics or central nervous system (CNS) involvement, underwent whole-genome sequencing (WGS) to identify structural variants (SVs) in the leukemic blasts. Unique sequences generated by SVs were targeted with patient-specific droplet digital PCR (ddPCR) assays. Genomic DNA (gDNA) from BM and cell-free DNA (cfDNA) from plasma and cerebrospinal fluid (CSF) were analyzed longitudinally. WGS with 30× coverage enabled target identification in all cases. Limit of quantifiability (LoQ) and limit of detection (LoD) for the ddPCR assays (n = 15) were up to 10−5 and 10−6, respectively. All targets were readily detectable in a multiplexed ddPCR with minimal DNA input (1 ng of gDNA) at a 10−1 dilution, and targets for half of the patients were also detectable at a 10−2 dilution. The level of MRD in BM at end of induction and end of consolidation block 1 was in a comparable range between ddPCR and clinical routine methods for samples with detectable residual disease, although our approach consistently detected higher MRD values for patients with B-cell precursor ALL. Additionally, several samples with undetectable MRD by flow cytometry were MRD-positive by ddPCR. In plasma, the level of leukemic targets decreased in cfDNA over time following the MRD level detected in BM. cfDNA was successfully extracted from all diagnostic CSF samples (n = 6), and leukemic targets were detected in half of these. The results suggest that our approach to design molecular assays, together with ddPCR quantification, is a technically feasible option for accurate MRD quantification and that cfDNA may contribute valuable information regarding MRD and low-grade CNS involvement.