Comparing three machine learning approaches to design a risk assessment tool for future fractures: predicting a subsequent major osteoporotic fracture in fracture patients with osteopenia and osteoporosis.

Four machine learning models were developed and compared to predict the risk of a future major osteoporotic fracture (MOF), defined as hip, wrist, spine and humerus fractures, in patients with a prior fracture. We developed a user-friendly tool for risk calculation of subsequent MOF in osteopenia patients, using the best performing model. INTRODUCTION: Major osteoporotic fractures (MOFs), defined as hip, wrist, spine and humerus fractures, can have serious consequences regarding morbidity and mortality. Machine learning provides new opportunities for fracture prediction and may aid in targeting preventive interventions to patients at risk of MOF. The primary objective is to develop and compare several models, capable of predicting the risk of MOF as a function of time in patients seen at the fracture and osteoporosis outpatient clinic (FO-clinic) after sustaining a fracture. METHODS: Patients aged > 50 years visiting an FO-clinic were included in this retrospective study. We compared discriminative ability (concordance index) for predicting the risk on MOF with a Cox regression, random survival forests (RSF) and an artificial neural network (ANN)-DeepSurv model. Missing data was imputed using multiple imputations by chained equations (MICE) or RSF's imputation function. Analyses were performed for the total cohort and a subset of osteopenia patients without vertebral fracture. RESULTS: A total of 7578 patients were included, 805 (11%) patients sustained a subsequent MOF. The highest concordance-index in the total dataset was 0.697 (0.664-0.730) for Cox regression; no significant difference was determined between the models. In the osteopenia subset, Cox regression outperformed RSF (p = 0.043 and p = 0.023) and ANN-DeepSurv (p = 0.043) with a c-index of 0.625 (0.562-0.689). Cox regression was used to develop a MOF risk calculator on this subset. CONCLUSION: We show that predicting the risk of MOF in patients who already sustained a fracture can be done with adequate discriminative performance. We developed a user-friendly tool for risk calculation of subsequent MOF in patients with osteopenia.

Quality of life after prostate cancer treatments in patients comparable at baseline.

BACKGROUND: Previous studies on the effects of different prostate cancer treatments on quality of life, were confounded because patients were not comparable. This study examined treatment effects in more comparable groups. METHODS: From 2008-2011, 240 patients with localised prostate cancer were selected to be eligible for both radical prostatectomy (RP) and external beam radiotherapy (EBRT). Brachytherapy (BT) was a third option for some. Health-related quality of life was measured by expanded prostate cancer index composite (EPIC) up to 12 months after treatment. RESULTS: In the sexual domain, RP led to worse summary scores (P<0.001) and more often to a clinically relevant deterioration from baseline than BT and EBRT (79%, 33%, 34%, respectively). In the urinary domain, RP also led to worse summary scores (P=0.014), and more deterioration from baseline (41%, 12%, 19%, respectively). Only on the irritative/obstructive urinary scale, more BT patients (40%) showed a relevant deterioration than RP (17%) and EBRT patients (11%). In the bowel domain, the treatment effects did not differ. CONCLUSION: This study provides a more unbiased comparison of treatment effects, as men were more comparable at baseline. Our results suggest that, for quality of life, radiotherapy is as least as good an option as RP for treating localised prostate cancer.

Non-protein-bound iron and free radical damage in fetuses with rhesus haemolytic disease: influence of intrauterine transfusions.

OBJECTIVE: To determine iron-induced free radical damage in fetal rhesus haemolytic disease (RHD) before and after repeated intrauterine red blood cell transfusions and its relation to hydrops fetalis. DESIGN: Prospective, observational study. SETTING: Department of Obstetrics, Leiden University Medical Centre, the Netherlands. POPULATION: Fifty anaemic fetuses, including 13 hydropic ones, 9 preterm and 12 term neonates and 8 female non-pregnant adults. METHODS: Venous blood plasma samples were collected from 50 fetuses suffering from RHD preliminary to the first, and if appropriate, subsequent intrauterine red blood cell transfusions for determination of iron status including non-protein-bound iron (NPBI) and iron-binding primary antioxidant proteins, total plasma anti-oxidant capacity and its contributing secondary antioxidants (e.g. vitamin C, uric acid, sulphydryl groups and peroxidation products). Results were compared with values of healthy preterm and term neonates directly at birth and adult controls. Within the fetal haemolytic group, 13 hydropic fetuses were analysed as a separate group. MAIN OUTCOME MEASURES: Non-protein-bound iron, antioxidants, total antioxidant capacity and peroxidation products. Sub analysis of the outcome measures of the hydropic fetuses. RESULTS: RHD fetuses had at initial cordocentesis a significantly higher NPBI level and a significantly lower total plasma antioxidant capacity than control babies and adults. Their vitamin C tended to be more oxidised but lipid peroxidation had not increased, when compared with preterm babies. The repeated intrauterine red blood cell transfusions had a positive effect on the total antioxidant capacity of plasma and did not increase the concentration of NPBI. The hydropic fetuses, who had higher NPBI concentrations and lower plasma protein concentrations and total antioxidant capacity, did not show more peroxidation products in plasma than the non-hydropic fetuses. Fetuses without reversal of hydrops despite intrauterine transfusions showed decreasing levels of proteins with subsequent transfusions but peroxidation products remained constant. CONCLUSION: Repeated intrauterine red blood cell transfusions do not lead to free radical damage in the fetus in utero. Iron-induced free radical activity does not appear to play a causative role in the proceeding of hydrops fetalis in RHD.

The capacity of different infusion fluids to lower the prooxidant activity of plasma iron: an important factor in resuscitation?

BACKGROUND: Prooxidant activity of non-protein-bound iron (NPBI) is an important contributor to reactive oxygen species-induced injury after the resuscitation of critically ill patients. Plasma NPBI occurs in critically ill adults, children, and newborn babies, who often require resuscitation. The ability of the resuscitation fluids to bind iron and lower the patients' NPBI levels in vitro has not previously been studied. STUDY DESIGN AND METHODS: In an in vitro model, highly iron-saturated cord blood plasma from 10 preterm and 10 term babies was mixed with FFP, pasteurized plasma protein solution, and 0.9-percent saline. Plasma from 10 healthy adult volunteers was used as a control. Before and after the mixing with any resuscitation fluid, NPBI levels and ceruloplasmin iron-oxidizing and transferrin iron-binding antioxidant capacities were measured. RESULTS: After the in vitro mixing with FFP, the incidence and concentration of NPBI were markedly decreased and the iron-binding antioxidant capacity was increased in the plasma of the preterm and term babies. Being mixed with pasteurized plasma protein solution and 0.9-percent saline did not influence the iron-binding antioxidant capacity of newborn babies' plasma. In the control plasma, results were not changed after the mixing with any resuscitation fluid. In every group, the iron-oxidizing antioxidant capacity was not changed after the mixing with any fluid. CONCLUSION: Iron-induced oxidative tissue damage may be influenced by resuscitation fluids. In the ongoing debate over the choice of crystalloid or colloid resuscitation fluids, the influence of each fluid on the patient's antioxidant capacity warrants more attention.

Cross-bridge kinetics in rat myocardium: effect of sarcomere length and calcium activation.

We tested the hypotheses that Ca(2+) concentration ([Ca(2+)]) and sarcomere length (SL) modulate force development via graded effects on cross-bridge kinetics in chemically permeabilized rat cardiac trabeculae. Using sinusoidal length perturbations, we derived the transfer functions of stiffness over a range of [Ca(2+)] at a constant SL of 2.1 micrometer (n = 8) and at SL of 2.0, 2.1, and 2.2 micrometer (n = 4). We found that changes in SL affected only the magnitude of stiffness, whereas [Ca(2+)] affected the magnitude and phase-frequency relations. The data were fit to complex functions of two exponential processes. The characteristic frequencies (b and c) of these processes are indexes of cross-bridge kinetics, with b relating to cross-bridge attachment to and c to detachment from certain non-force-generating states. Both were significantly affected by [Ca(2+)], with an increase in b and c of 140 and 44%, respectively, over the range of [Ca(2+)] studied (P < 0.01). In contrast, SL had no effect on the characteristic frequencies (P > 0.6). We conclude that Ca(2+) activation modulates force development in rat myocardium, at least in part, via a graded effect on cross-bridge kinetics, whereas SL effects are mediated mainly by recruitment of cross bridges.

The organization of leg movements in preterm and full-term infants after term age.

In a sample of 13 full-term and 10 preterm infants, the development of kicking movements was studied at 6, 12, and 18 weeks (corrected) age. In healthy full-term infants some characteristics are strikingly stable, such as the duration of the flexion and extension phase and the within-joint organization. These parameters did not differ in preterm compared to full-term infants. For other features, however, developmental changes and differences were observed. Full-term infants tended to decrease their kick frequencies slightly with age. In preterm infants much higher initial kick rates were found, followed by a steep decrease, which resulted in kick frequencies comparable to the full-term levels after the (corrected) age of 12 weeks. There is a tight coupling between the movements in the different joints of the leg in full-term newborns. Preterm infants, in contrast, initially show much lower cross-correlations between hip and ankle and between knee and ankle. This is particularly the case for those preterm infants who were born before 32 weeks gestation. Again, the differences resolved after the age of 12 weeks, which might be related to a transformation in neural functions reported previously around this age. The initial differences in the characteristics of kicking appeared to be more readily explainable by differences in neurological condition than by contrasts in leg volume or postural control.

The development of head position preference in preterm infants beyond term age.

Healthy full-term infants show a developmental trend in head position from an initial right-sided preference to one with the head in midline around the age of 12 weeks. We studied the effects of intrauterine growth retardation (IUGR) and the degree of prematurity on both aspects of development from 35 weeks postmenstrual age to 18 weeks corrected age in 35 preterm infants without overt neurological abnormalities and whose gestational ages ranged from 27 to 34 weeks. Our data reveal that, during the preterm period, infants born after pregnancies of 32 weeks or less showed a lack of right-sided preferences for head turning after release from midline but not for the subsequent maintenance of a position. IUGR did not seem to affect either preference. After term age a right-sided preference diminished while a head midline position increased. The latter was not significantly delayed in relation to birth before 32 weeks gestation or IUGR. However, when infants were classified on the basis of neurological differences as reflected in a (mildly) abnormal movement quality, a delay in the attainment of a midline posture was observed, which suggests it is related to a suboptimal neurological condition. This delay, however, was also accounted for by the side-to-side flattening of the skull.

Qualitative changes in general movements and their prognostic value in preterm infants.

Qualitative assessments of general movements have been shown to have considerable power in predicting neurological outcomes in preterm infants with brain damage. In the present study such assessments were made in 35 preterm infants without major neurological problems before term age, born between 27 and 34 weeks gestation, of whom 12 were small-for-gestational-age (SGA). Most infants maintained a normal or (mildly) abnormal quality from 35 weeks postmenstrual age through 6, 12, 18 to 24 weeks corrected age. Seven changed from initially abnormal movements to a normal quality, six of them after 12 weeks. Differences between SGA and appropriate-for-gestational-age infants became less evident with age, particularly after 12 weeks. This was not the case when comparisons were made on the basis of gestational ages below or above 32 weeks. The prediction of neurological and mental outcomes at 1 year also improved after 12 weeks, around which age a major transformation in neural functions occurs. It is concluded that assessments of movement quality are particularly successful in predicting abnormal outcomes in comparison to examinations based on muscle tone and elicited responses.

Effects of birthweight status and gestational age on the quality of general movements in preterm newborns.

It is hypothesised that individual differences in nervous system functioning, undetected by a neurological examination, are reflected in the quality of spontaneous movements of preterm newborns. Given this hypothesis it is expected that a short pregnancy duration and IUGR will be related to an abnormal movement quality. These expectations were confirmed in a group of 37 small-for-gestational-age and appropriate-for-gestational-age preterm newborns with gestational ages ranging from 27 to 34 weeks without serious perinatal complications and for whom no overt neurological abnormalities could be detected based on the evaluation of elicited responses and tonus at the postmenstrual age of 35 weeks. The quality of general movements was adversely affected by both IUGR and a pregnancy duration below 32 weeks. Newborns with an abnormal movement quality also had significantly lower obstetrical optimality scores. Previous research has shown these scores to be related to the neurological condition of the newborn. We conclude that observations of movement quality, being neither intrusive nor time consuming, may constitute a useful addition to the neurological assessment of preterm newborns without serious perinatal complications.

Hyperventilation-induced changes in periodic oscillations in forehead skin blood flow measured by laser Doppler flowmetry.

Rhythmic oscillations in forehead skin blood flow were studied with the laser Doppler technique in thirteen healthy subjects. During voluntary hyperventilation, a three-fold increase in relative amplitude of the spontaneous rhythmic oscillations in forehead skin blood flow was observed, whereas mean blood flow decreased by 15%. During hyperventilation, the relative amplitude of the oscillations was on average 36% of the mean blood flow value. The mean incidence of the oscillations increased significantly, from 68% of the measuring time before, to 96% of the measuring time during hyperventilation. The oscillation frequency was not affected by hyperventilation. Before, during and after hyperventilation the average oscillation frequency was 0.140 Hz (8.4 min-1), 0.145 Hz (8.7 min-1) respectively. The application of heat or a local anaesthetic to the skin attenuated the relative amplitude of the oscillations in forehead skin blood flow during hyperventilation as well as before and after.

The 'metabolic burst' in polymorphonuclear leukocytes from patients with quiescent psoriasis.

We report an investigation of peripheral blood polymorphonuclear leukocytes (PMN) from untreated patients with mild, quiescent psoriasis. Four aspects of the 'metabolic burst' were measured to illustrate phagocytosis-related events. These were: myeloperoxidase activity, hydrogen peroxide release, superoxide production and luminol-amplified chemiluminescence. Our study does not support the concept of an intrinsic abnormality of the PMN in psoriasis, but provides additional evidence for disease activity-dependent changes in phagocytic behaviour. Possible interactions with certain humoral factors, the so-called opsonins, are discussed.

Monocyte function is normal in quiescent psoriasis.

We report an investigation of peripheral blood monocytes from untreated patients with mild, quiescent psoriasis. Possible metabolic changes were monitored by the determination of 3 enzymes representing different pathways of glucose metabolism and 2 lysosomal enzymes. Signal processing was evaluated by the measurement of cyclic AMP levels before and after hormonal stimulation. Luminol-amplified chemiluminescence provided an objective approach to assessing phagocytic capacity. Finally, the pattern of maturation of normal and psoriatic monocytes has been compared during culture in vitro. Our results were uniformly and wholly negative; we conclude that the concept of an "intrinsic" abnormality of the psoriatic monocyte may be excluded. Possible reasons for discrepancies in the literature are discussed.