Exploratory Analysis of Methods for Automated Classification of Laboratory Test Orders into Syndromic Groups in Veterinary Medicine

Background

Recent focus on earlier detection of pathogen introduction in human and animal populations has led to the development of surveillance systems based on automated monitoring of health data. Real- or near real-time monitoring of pre-diagnostic data requires automated classification of records into syndromes–syndromic surveillance–using algorithms that incorporate medical knowledge in a reliable and efficient way, while remaining comprehensible to end users.

Methods

This paper describes the application of two of machine learning (Naïve Bayes and Decision Trees) and rule-based methods to extract syndromic information from laboratory test requests submitted to a veterinary diagnostic laboratory.

Results

High performance (F₁-macro = 0.9995) was achieved through the use of a rule-based syndrome classifier, based on rule induction followed by manual modification during the construction phase, which also resulted in clear interpretability of the resulting classification process. An unmodified rule induction algorithm achieved an F_1-micro score of 0.979 though this fell to 0.677 when performance for individual classes was averaged in an unweighted manner (F_1-macro), due to the fact that the algorithm failed to learn 3 of the 16 classes from the training set. Decision Trees showed equal interpretability to the rule-based approaches, but achieved an F_1-micro score of 0.923 (falling to 0.311 when classes are given equal weight). A Naïve Bayes classifier learned all classes and achieved high performance (F_1-micro = 0.994 and F_1-macro = .955), however the classification process is not transparent to the domain experts.

Conclusion

The use of a manually customised rule set allowed for the development of a system for classification of laboratory tests into syndromic groups with very high performance, and high interpretability by the domain experts. Further research is required to develop internal validation rules in order to establish automated methods to update model rules without user input.     

Differential Contributions of Vision,Touch and Muscle Proprioception to the Coding of Hand Movements

To further elucidate the mechanisms underlying multisensory integration, this study examines the controversial issue of whether congruent inputs from three different sensory sources can enhance the perception of hand movement. Illusory sensations of clockwise rotations of the right hand were induced by either separately or simultaneously stimulating visual, tactile and muscle proprioceptive channels at various intensity levels. For this purpose, mechanical vibrations were applied to the pollicis longus muscle group in the subjects’ wrists, and a textured disk was rotated under the palmar skin of the subjects’ right hands while a background visual scene was projected onto the rotating disk. The elicited kinaesthetic illusions were copied by the subjects in real time and the EMG activity in the adductor and abductor wrist muscles was recorded. The results show that the velocity of the perceived movements and the amplitude of the corresponding motor responses were modulated by the nature and intensity of the stimulation. Combining two sensory modalities resulted in faster movement illusions, except for the case of visuo-tactile co-stimulation. When a third sensory input was added to the bimodal combinations, the perceptual responses increased only when a muscle proprioceptive stimulation was added to a visuo-tactile combination. Otherwise, trisensory stimulation did not override bimodal conditions that already included a muscle proprioceptive stimulation. We confirmed that vision or touch alone can encode the kinematic parameters of hand movement, as is known for muscle proprioception. When these three sensory modalities are available, they contribute unequally to kinaesthesia. In addition to muscle proprioception, the complementary kinaesthetic content of visual or tactile inputs may optimize the velocity estimation of an on-going movement, whereas the redundant kinaesthetic content of the visual and tactile inputs may rather enhance the latency of the perception.     

Modelling the cost-effectiveness of mass screening and treatment for reducing Plasmodium falciparum malaria burden

Background

Past experience and modelling suggest that, in most cases, mass treatment strategies are not likely to succeed in interrupting Plasmodium falciparum malaria transmission. However, this does not preclude their use to reduce disease burden. Mass screening and treatment (MSAT) is preferred to mass drug administration (MDA), as the latter involves massive over-use of drugs. This paper reports simulations of the incremental cost-effectiveness of well-conducted MSAT campaigns as a strategy for P. falciparum malaria disease-burden reduction in settings with varying receptivity (ability of the combined vector population in a setting to transmit disease) and access to case management.

Methods

MSAT incremental cost-effectiveness ratios (ICERs) were estimated in different sub-Saharan African settings using simulation models of the dynamics of malaria and a literature-based MSAT cost estimate. Imported infections were simulated at a rate of two per 1,000 population per annum. These estimates were compared to the ICERs of scaling up case management or insecticide-treated net (ITN) coverage in each baseline health system, in the absence of MSAT.

Results

MSAT averted most episodes, and resulted in the lowest ICERs, in settings with a moderate level of disease burden. At a low pre-intervention entomological inoculation rate (EIR) of two infectious bites per adult per annum (IBPAPA) MSAT was never more cost-effective than scaling up ITNs or case management coverage. However, at pre-intervention entomological inoculation rates (EIRs) of 20 and 50 IBPAPA and ITN coverage levels of 40 or 60%, respectively, the ICER of MSAT was similar to that of scaling up ITN coverage further.

Conclusions

In all the transmission settings considered, achieving a minimal level of ITN coverage is a “best buy”. At low transmission, MSAT probably is not worth considering. Instead, MSAT may be suitable at medium to high levels of transmission and at moderate ITN coverage. If undertaken as a burden-reducing intervention, MSAT should be continued indefinitely and should complement, not replace, case management and vector control interventions.     

Functionality of lactic acid bacteria peptidase activities in the hydrolysis of gliadin‐like fragments

Aims: To evaluate the role of the peptidase activities from sourdough lactic acid bacteria (LAB) in the degradation of α‐gliadin fragments. Methods and Results: Different proline‐containing substrates were hydrolysed by LAB indicating pro‐specific peptidase activities. Lactobacillus plantarum CRL 775 and Pediococcus pentosaceus CRL 792 displayed the highest tri‐ and di‐peptidase activities, respectively. Lactobacillus plantarum strains hydrolysed more than 60%α‐gliadin fragments corresponding to the 31–43 and 62–75 amino acids in the protein after 2 h. None of the LAB strains alone could hydrolyse 57–89 α‐gliadin peptide; however, the combination of L. plantarum CRL 775 and P. pentosaceus CRL 792 led to hydrolysis (57%) of this peptide in 8 h. Conclusions: The capacity of LAB strains to degrade α‐gliadin fragments was not correlated to individual peptidase activities. Several strains separately degraded the 31–43 and 62–75 α‐gliadin fragments, while the 57–89 peptide degradation was associated with the combination of peptidase profiles from pooled LAB strains. This is the first report on the peptide hydrolase system of sourdough pediococci and its ability to reduce α‐gliadin fragments. Significance and Impact of the Study: This study contributes to a better knowledge of sourdough LAB proteolytic system and its role in the degradation of proline‐rich α‐gliadin peptides involved in celiac disease.     

Hybrid elementary flux analysis/nonparametric modeling: application for bioprocess control

Background  

The progress in the "-omic" sciences has allowed a deeper knowledge on many biological systems with industrial interest. This knowledge is still rarely used for advanced bioprocess monitoring and control at the bioreactor level. In this work, a bioprocess control method is presented, which is designed on the basis of the metabolic network of the organism under consideration. The bioprocess dynamics are formulated using hybrid rigorous/data driven systems and its inherent structure is defined by the metabolism elementary modes.     

SIRAC: Supervised Identification of Regions of Aberration in aCGH datasets

Background  

Array comparative genome hybridization (aCGH) provides information about genomic aberrations. Alterations in the DNA copy number may cause the cell to malfunction, leading to cancer. Therefore, the identification of DNA amplifications or deletions across tumors may reveal key genes involved in cancer and improve our understanding of the underlying biological processes associated with the disease.