How to automatically turn patient experience free-text responses into actionable insights: a natural language programming (NLP) approach.

BACKGROUND: Patient experience surveys often include free-text responses. Analysis of these responses is time-consuming and often underutilized. This study examined whether Natural Language Processing (NLP) techniques could provide a data-driven, hospital-independent solution to indicate points for quality improvement. METHODS: This retrospective study used routinely collected patient experience data from two hospitals. A data-driven NLP approach was used. Free-text responses were categorized into topics, subtopics (i.e. n-grams) and labelled with a sentiment score. The indicator 'impact', combining sentiment and frequency, was calculated to reveal topics to improve, monitor or celebrate. The topic modelling architecture was tested on data from a second hospital to examine whether the architecture is transferable to another hospital. RESULTS: A total of 38,664 survey responses from the first hospital resulted in 127 topics and 294 n-grams. The indicator 'impact' revealed n-grams to celebrate (15.3%), improve (8.8%), and monitor (16.7%). For hospital 2, a similar percentage of free-text responses could be labelled with a topic and n-grams. Between-hospitals, most topics (69.7%) were similar, but 32.2% of topics for hospital 1 and 29.0% of topics for hospital 2 were unique. CONCLUSIONS: In both hospitals, NLP techniques could be used to categorize patient experience free-text responses into topics, sentiment labels and to define priorities for improvement. The model's architecture was shown to be hospital-specific as it was able to discover new topics for the second hospital. These methods should be considered for future patient experience analyses to make better use of this valuable source of information.

Soil nutrient status determines how elephant utilize trees and shape environments.

1. Elucidation of the mechanism determining the spatial scale of patch selection by herbivores has been complicated by the way in which resource availability at a specific scale is measured and by vigilance behaviour of the herbivores themselves. To reduce these complications, we studied patch selection by an animal with negligible predation risk, the African elephant. 2. We introduce the concept of nutrient load as the product of patch size, number of patches and local patch nutrient concentration. Nutrient load provides a novel spatially explicit expression of the total available nutrients a herbivore can select from. 3. We hypothesized that elephant would select nutrient-rich patches, based on the nutrient load per 2500 m(2) down to the individual plant scale, and that this selection will depend on the nitrogen and phosphorous contents of plants. 4. We predicted that elephant would cause more adverse impact to trees of lower value to them in order to reach plant parts with higher nutrient concentrations such as bark and root. However, elephant should maintain nutrient-rich trees by inducing coppicing of trees through re-utilization of leaves. 5. Elephant patch selection was measured in a homogenous tree species stand by manipulating the spatial distribution of soil nutrients in a large field experiment using NPK fertilizer. 6. Elephant were able to select nutrient-rich patches and utilized Colophospermum mopane trees inside these patches more than outside, at scales ranging from 2500 down to 100 m(2) . 7. Although both nitrogen and phosphorus contents of leaves from C. mopane trees were higher in fertilized and selected patches, patch choice correlated most strongly with nitrogen content. As predicted, stripping of leaves occurred more in nutrient-rich patches, while adverse impact such as uprooting of trees occurred more in nutrient-poor areas. 8. Our results emphasize the necessity of including scale-dependent selectivity in foraging studies and how elephant foraging behaviour can be used as indicators of change in the availability of nutrients.

Circulatory model of vascular and interstitial distribution kinetics of rocuronium: a population analysis in patients.

The time-course of the neuromuscular blocking effect of rocuronium depends on circulatory mixing and the rate of distribution into the interstitial space. In order to quantitatively evaluate these processes, a physiologically meaningful model of distribution kinetics based on circulatory transport and interstitial diffusion, was fitted to rocuronium disposition data in 10 patients using a population approach. Information on cardiac output and circulatory mixing was obtained from the kinetics of indocyanine green (ICG), which was injected simultaneously with rocuronium. As a compromise between physiological reality and parameter identifiability, the organs of the systemic circulation were lumped into a heterogeneous subsystem, described by an axially distributed model of extravascular diffusion. Diffusion into the interstitial space determines the rate of rocuronium distribution in the body (diffusional time constant 89 min). The resulting whole body distribution kinetics depends both on cardiac output and on the apparent permeability surface area product (0.16 l/min). The analysis of the ICG data revealed that heterogeneity of blood transit time through the systemic circulation decreased and that cardiopulmonary volume increased, respectively, with cardiac output. The approach should be useful for studying the effect of disease states on distribution kinetics of drugs.

Pulse dye densitometry and indocyanine green plasma disappearance in ASA physical status I-II patients.

BACKGROUND: Indocyanine green plasma disappearance rate (ICG-PDR) is used to evaluate hepatic function. Although hepatic failure is generally said to occur with an ICG-PDR <18%/min, ICG disappearance rate is poorly defined in the healthy population, and a clear cutoff value of ICG-PDR that discriminates between normal hepatic function and hepatic failure has not yet been described. We therefore defined the ICG disappearance rate in an otherwise healthy patient population. In addition, we evaluated the noninvasive measurement of ICG-PDR (transcutaneously by pulse dye densitometry [PDD] at the finger and the nose) and compared these with the simultaneously performed invasive measurements of ICG-PDR (in arterial blood). METHODS: In patients without signs of liver disease, scheduled for elective nonhepatic surgery, 10 mg ICG was administered IV and ICG-PDR measured by PDD (DDG-2001, Nihon Kohden, Tokyo, Japan). In a subset of patients, arterial blood samples were gathered to compare PDD with invasive ICG measurements. Methods were compared using Bland-Altman analysis. The results of our study and reported studies on discriminative use of ICG-PDR in assessing liver failure were used to construct receiver operating characteristic curves. RESULTS: Forty-one patients were studied: 33 using the finger probe and 8 using the nose probe. The mean +/- SD noninvasive ICG-PDR in this patient population is 23.1% +/- 7.9%/min (n = 41) with a range of 9.7% to 43.2%/min. Bias (+/-2 sd, limits of agreement) for ICG-PDR measured by PDD compared with those measured in arterial blood were 1.6%/min (-5.2% to 8.3%/min) for the finger probe and -6.0%/min (-15.5% to 3.4%/min) for the nose probe. CONCLUSION: ICG-PDR values in a population without liver failure ranged well below 18%/min, cited as the cutoff value for hepatic failure. This cutoff value needs reconsideration. In addition, we conclude that the ICG concentration is adequately determined noninvasively by PDD.

Cardiovascular monitoring by pulse dye densitometry or arterial indocyanine green dilution.

BACKGROUND: Noninvasive cardiac output (CO) monitoring is possible by indocyanine green (ICG) dilution measured by pulse dye densitometry (PDD). To validate the precision of this method, we compared hemodynamic variables derived from PDD (DDG-2001, Nihon Kohden, Japan) with those derived from simultaneously taken arterial blood ICG concentrations. METHODS: In 20 patients (6 M/14 F), ASA I or II, 36 sessions were performed (n = 24 with the PDD-finger probe, n = 10 with the PDD-nose probe). After IV administration of 10 mg ICG, 34 arterial blood samples were taken during each session, with 20 samples taken during the first 2 min. CO, central blood volume (CBV), and total blood volume (TBV) were calculated independently from ICG and PDD and the results compared between methods using Bland-Altman analysis. The results are reported as mean difference (bias) and limits of agreement (LOA = +/- 2 sd). RESULTS: PDD using the finger probe underestimated CO (LOA) by 5% (-56% and 47%); overestimated CBV by 21% (-54% and 96%) and underestimated TBV by -15% (-38% and 8%). PDD using the nose probe overestimated CO (LOA) by 30% (-67% and 127%); CBV by 48% (-98% and 193%) and underestimated TBV by -10% (-47% and 27%). CONCLUSION: Despite the permissible bias, the wide LOA of the PDD-derived hemodynamic variables CO and CBV, compared with those simultaneously obtained by invasive arterial ICG measurements, suggest that PDD is unsuitable for evaluation of cardiovascular variables in the individual patient. Hence, the reliability and clinical use of this method seem limited.

Cardiovascular parameters and liver blood flow after infusion of a colloid solution and epidural administration of ropivacaine 0.75%: the influence of age and level of analgesia.

BACKGROUND AND OBJECTIVE: Lumbar epidural anaesthesia induces cardiovascular changes and decreases liver blood flow (Qh). We studied the effects of age on haemodynamics, blood volumes and Qh before and after epidural anaesthesia. METHODS: Thirty-six patients were enrolled as follows: group 1, 20-44 years; group 2, 45-70 years; group 3, >70 years. Using pulse dye densitometry, in addition to heart rate and arterial blood pressure (arterial BP), cardiac output, total blood volume, central blood volume and Qh were measured, before and after colloid infusion (500 ml hydroxyethyl starch, 6%) and after epidural administration of 15 ml of 0.75% ropivacaine. RESULTS: With age the level of analgesia [median (range)] increased from T7 (L2-T4) in group 1 to T4 (T10-C7) in group 3 (P = 0.04). After colloid infusion, heart rate (mean difference +/- SE; 2.1 +/- 0.7 beats min(-1)), systolic BP (4.1 +/- 2.2 mmHg) and Qh 162 ml min(-1) (ratio 0.90, 95% confidence interval 0.81-0.99) increased slightly but significantly, and were unaffected by age. Epidural anaesthesia induced a significant decrease in Qh (265 ml min(-1); ratio 1.20, 95% confidence interval 1.07-1.35) and arterial pressure (for systolic BP: P = 1 x 10(-7)). A significantly larger decrease in systolic BP occurred in the older, compared with the middle, age group (P = 0.04). Age did not affect epidural-induced changes in cardiac output, total and central blood volumes, and Qh. CONCLUSION: Age increases the level of analgesia after epidural ropivacaine and is associated with a more pronounced decrease in arterial pressure. A colloid preload mildly increases haemodynamics, but this insufficiently prevents younger and elderly patients from a decrease in Qh after lumbar epidural anaesthesia.