Detection of allergic transfusion-related adverse events from electronic medical records.

BACKGROUND: Transfusion-related adverse events can be unrecognized and unreported. As part of the US Food and Drug Administration's Center for Biologics Evaluation and Research Biologics Effectiveness and Safety initiative, we explored whether machine learning methods, such as natural language processing (NLP), can identify and report transfusion allergic reactions (ARs) from electronic health records (EHRs). STUDY DESIGN AND METHODS: In a 4-year period, all 146 reported transfusion ARs were pulled from a database of 86,764 transfusions in an academic health system, along with a random sample of 605 transfusions without reported ARs. Structured and unstructured EHR data were retrieved, including demographics, new symptoms, medications, and lab results. In unstructured data, evidence from clinicians' notes, test results, and prescriptions fields identified transfusion ARs, which were used to extract NLP features. Clinician reviews of selected validation cases assessed and confirmed model performance. RESULTS: Clinician reviews of selected validation cases yielded a sensitivity of 67.9% and a specificity of 97.5% at a threshold of 0.9, with a positive predictive value (PPV) of 84%, estimated to 4.5% when extrapolated to match transfusion AR incidence in the full transfusion dataset. A higher threshold achieved sensitivity of 43% with specificity/PPV of 100% in our validation set. Essential features predicting ARs were recognized transfusion reactions, administration of antihistamines or glucocorticoids, and skin symptoms (e.g., hives and itching). Removal of NLP features decreased model performance. DISCUSSION: NLP algorithms can identify transfusion reactions from the EHR with a reasonable level of precision for subsequent clinician review and confirmation.

Automated Microfluidic Platform for Serial Polymerase Chain Reaction and High-Resolution Melting Analysis.

We report the development of an automated genetic analyzer for human sample testing based on microfluidic rapid polymerase chain reaction (PCR) with high-resolution melting analysis (HRMA). The integrated DNA microfluidic cartridge was used on a platform designed with a robotic pipettor system that works by sequentially picking up different test solutions from a 384-well plate, mixing them in the tips, and delivering mixed fluids to the DNA cartridge. A novel image feedback flow control system based on a Canon 5D Mark II digital camera was developed for controlling fluid movement through a complex microfluidic branching network without the use of valves. The same camera was used for measuring the high-resolution melt curve of DNA amplicons that were generated in the microfluidic chip. Owing to fast heating and cooling as well as sensitive temperature measurement in the microfluidic channels, the time frame for PCR and HRMA was dramatically reduced from hours to minutes. Preliminary testing results demonstrated that rapid serial PCR and HRMA are possible while still achieving high data quality that is suitable for human sample testing.

Automated Classification and Cluster Visualization of Genotypes Derived from High Resolution Melt Curves.

INTRODUCTION: High Resolution Melting (HRM) following PCR has been used to identify DNA genotypes. Fluorescent dyes bounded to double strand DNA lose their fluorescence with increasing temperature, yielding different signatures for different genotypes. Recent software tools have been made available to aid in the distinction of different genotypes, but they are not fully automated, used only for research purposes, or require some level of interaction or confirmation from an analyst. MATERIALS AND METHODS: We describe a fully automated machine learning software algorithm that classifies unknown genotypes. Dynamic melt curves are transformed to multidimensional clusters of points whereby a training set is used to establish the distribution of genotype clusters. Subsequently, probabilistic and statistical methods were used to classify the genotypes of unknown DNA samples on 4 different assays (40 VKORC1, CYP2C9*2, CYP2C9*3 samples in triplicate, and 49 MTHFR c.665C>T samples in triplicate) run on the Roche LC480. Melt curves of each of the triplicates were genotyped separately. RESULTS: Automated genotyping called 100% of VKORC1, CYP2C9*3 and MTHFR c.665C>T samples correctly. 97.5% of CYP2C9*2 melt curves were genotyped correctly with the remaining 2.5% given a no call due to the inability to decipher 3 melt curves in close proximity as either homozygous mutant or wild-type with greater than 99.5% posterior probability. CONCLUSIONS: We demonstrate the ability to fully automate DNA genotyping from HRM curves systematically and accurately without requiring any user interpretation or interaction with the data. Visualization of genotype clusters and quantification of the expected misclassification rate is also available to provide feedback to assay scientists and engineers as changes are made to the assay or instrument.

Identification and comparison of insulin pharmacokinetics injected with a new 4-mm needle vs 6- and 8-mm needles accounting for endogenous insulin and C-peptide secretion kinetics in non-diabetic adult males.

AIMS/INTRODUCTION: Many patients with diabetes now use 5-, 6- or 8-mm needles for insulin injection. However, it is unclear whether needle length, particularly for shorter needles, affects the pharmacokinetic properties of insulin. MATERIALS AND METHODS: This was a three-way, randomized, cross-over, single-center study involving 12 healthy Japanese adult males (age 27.4 ± 4.14 years; weight 64.2 ± 5.2 kg; body fat percentage 18.2 ± 1.5%). Participants received a subcutaneous (abdomen) dose of insulin lispro (1.5 U for participants weighing 55 to <65.0 kg; 2.0 U for participants weighing 65.0 to <80.0 kg) delivered using a 32-G × 4 mm (32G × 4), 31-G × 8 mm (31G × 8) or 32-G × 6 mm (32G × 6) needle with a 3-7-day washout between doses. Pharmacokinetic parameters of exogenous insulin were identified using non-linear least squares, where the total insulin concentration was fit to the measured plasma insulin concentration using an overall combined model that accounted for C-peptide/insulin secretion in addition to the injected dose. RESULTS: Maximum concentration and area under the curve for 0 to infinity min for insulin were bioequivalent for the 32G × 4 needle relative to the 32G × 6 and the 31G × 8 needles. The time to the maximum insulin concentration was bioequivalent for the 32G × 4 needle relative to the 32G × 6 needle, but not the 31G × 8 needle. CONCLUSIONS: The use of 4-mm needles is unlikely to change the pharmacokinetic properties of insulin when injected subcutaneously in adults. This trial was registered with UMIN-CTR (no. UMIN000004469).

The identifiable virtual patient model: comparison of simulation and clinical closed-loop study results.

BACKGROUND: Optimizing a closed-loop insulin delivery algorithm for individuals with type 1 diabetes can be potentially facilitated by a mathematical model of the patient. However, model simulation studies that evaluate changes to the control algorithm need to produce conclusions similar to those that would be obtained from a clinical study evaluating the same modification. We evaluated the ability of a low-order identifiable virtual patient (IVP) model to achieve this goal. METHODS: Ten adult subjects (42.5 ± 11.5 years of age; 18.0 ± 13.5 years diabetes; 6.9 ± 0.8% hemoglobin A1c) previously characterized with the IVP model were studied following the procedures independently reported in a pediatric study assessing proportional-integral-derivative control with and without a 50% meal insulin bolus. Peak postprandial glucose levels with and without the meal bolus and use of supplemental carbohydrate to treat hypoglycemia were compared using two-way analysis of variance and chi-square tests, respectively. RESULTS: The meal bolus decreased the peak postprandial glucose levels in both the adult-simulation and pediatricclinical study (231 ± 38 standard deviation to 205 ± 33 mg/dl and 226 ± 51 to 194 ± 47 mg/dl, respectively; p = .0472). No differences were observed between the peak postprandial levels obtained in the two studies (clinical and simulation study not different, p = .57; interaction p = .83) or in the use of supplemental carbohydrate (3 occurrences in 17 patient days of closed-loop control in the clinical-pediatric study; 7 occurrences over 20 patient days in the adult-simulation study, p = .29). CONCLUSIONS: Closed-loop simulations using an IVP model can predict clinical study outcomes in patients studied independently from those used to develop the model.

Identification of intraday metabolic profiles during closed-loop glucose control in individuals with type 1 diabetes.

BACKGROUND: Algorithms for closed-loop insulin delivery can be designed and tuned empirically; however, a metabolic model that is predictive of clinical study results can potentially accelerate the process. METHODS: Using data from a previously conducted closed-loop insulin delivery study, existing models of meal carbohydrate appearance, insulin pharmacokinetics, and the effect on glucose metabolism were identified for each of the 10 subjects studied. Insulin's effects to increase glucose uptake and decrease endogenous glucose production were described by the Bergman minimal model, and compartmental models were used to describe the pharmacokinetics of subcutaneous insulin absorption and glucose appearance following meals. The composite model, comprised of only five equations and eight parameters, was identified with and without intraday variance in insulin sensitivity (S(I)), glucose effectiveness at zero insulin (GEZI), and endogenous glucose production (EGP) at zero insulin. RESULTS: Substantial intraday variation in SI, GEZI and EGP was observed in 7 of 10 subjects (root mean square error in model fit greater than 25 mg/dl with fixed parameters and nadir and/or peak glucose levels differing more than 25 mg/dl from model predictions). With intraday variation in these three parameters, plasma glucose and insulin were well fit by the model (R(2) = 0.933 +/- 0.00971 [mean +/- standard error of the mean] ranging from 0.879-0.974 for glucose; R(2) = 0.879 +/- 0.0151, range 0.819-0.972 for insulin). Once subject parameters were identified, the original study could be reconstructed using only the initial glucose value and basal insulin rate at the time closed loop was initiated together with meal carbohydrate information (glucose, R(2) = 0.900 +/- 0.015; insulin delivery, R(2) = 0.640 +/- 0.034; and insulin concentration, R(2) = 0.717 +/- 0.041). CONCLUSION: Metabolic models used in developing and comparing closed-loop insulin delivery algorithms will need to explicitly describe intraday variation in metabolic parameters, but the model itself need not be comprised by a large number of compartments or differential equations.

Modeling insulin action for development of a closed-loop artificial pancreas.

Three models of glucose homeostasis are compared in terms of their steady-state dose-response characteristics, how they characterize glucose distribution kinetics, and how they characterize the dynamics of insulin action. The three models [minimal model, AIDA (Automated Insulin Dosage Advisor), and a model by Sorensen] are used to discuss a wider variety of questions related to metabolic modeling. Simulations are performed comparing each model's response to an intravenous glucose tolerance test, with and without incremental insulin responses, to existing data in individuals with type 1 diabetes mellitus. Predicted changes in blood glucose following a subcutaneous bolus of insulin or an incremental increase in basal insulin delivery are simulated. From these results, the models are evaluated as potential candidates for simulating changes in treatment and developing a closed-loop insulin delivery algorithm. While no consensus model is proposed, relevant issues needing to be addressed are highlighted.