Real-time near infrared artificial intelligence using scalable non-expert crowdsourcing in colorectal surgery

Surgical artificial intelligence (AI) has the potential to improve patient safety and clinical outcomes. To date, training such AI models to identify tissue anatomy requires annotations by expensive and rate-limiting surgical domain experts. Herein, we demonstrate and validate a methodology to obtain high quality surgical tissue annotations through crowdsourcing of non-experts, and real-time deployment of multimodal surgical anatomy AI model in colorectal surgery.

Detection of Botulinal Neurotoxins A, B, E, and F by Amplified Enzyme-Linked Immunosorbent Assay: Collaborative Study

An amplified enzyme-linked immunosorbent assay (amp-ELISA) was compared with the AOAC Official Method 977.26 for detection of Clostridium botulinum and its toxins in foods. Eleven laboratories participated and the results of 10 laboratories were used in the study. Two anaerobic culture media, tryptone peptone glucose yeast extract (TPGY) and cooked meat medium (CMM) were used to generate toxic samples with types A, B, E, and F botulinal strains. Nonbotulinal clostridia were also tested. The toxicity of each botulinal culture was determined by the AOAC method, and the cultures were then diluted, if necessary, to high (about 10 000 minimal lethal dose [MLD]/mL) and low (about 100 MLD/mL) test samples. The overall sensitivity of detection in TPGY and CMM cultures with the amp-ELISA was 94.7% at about 100 MLD/mL and 99.6% for samples with ≥10 000 MLD/mL toxicity. The amp-ELISA detection sensitivity for low toxin samples was 92.3% in TPGY and 99.4% in CMM. The false-positive rate ranged from 1.5% for type A to 28.6% for type F in TPGY, and from 2.4% for type A to 11.4% for type F in CMM. Most of the cross-reactivity was due to detection of other botulinal types, especially in high toxin samples. The amp-ELISA could be used to screen suspect cultures for botulinal toxins. Positive amp-ELISA samples would be confirmed by the AOAC reference method.

Quantitative determination of haptoglobin (HAP) in human and bovine sera by capillary zone electrophoresis (CZE)

This study describes an original assay for serum haptoglobin determination by measuring the capacity of human haptoglobin (hHAP) and bovine haptoglobin (bHAP) to bind haemoglobin (Hb) as established by capillary zone electrophoresis (CZE). This method involves the addition of Hb in excess to the serum and the separation of the HAP-Hb complexes from free haemoglobin by CZE. Protein migration was recorded at a wavelength of 415 nm which reveals Hb alone (free or bound), and the concentration of HAP was indirectly estimated by measuring bound Hb. Different CZE conditions and the peak migration time of Hb from various species (human, equine, bovine, canine) were investigated. The electrophoretic separation of free human Hb (hHb) in excess and the hHAP-hHb complex was fully achieved by CZE, allowing a quantitative determination of hHAP. However, bovine haemoglobin (bHb) bound to bHAP and free bHb were poorly separated under the same conditions. The best detachment between bHAP-Hb complexes and free Hb was only attained in the bovine sample by use of canine haemoglobin (cHb). CZE assays performed with cHb gave very close values to those of a classic photometric method which measured the peroxidase activity of the haptoglobin-cyanmethaemoglobin complexes (y = 1.0168x - 0.072; r2 = 0.97). CZE assay was fast (< 10 min), inexpensive, did not require the use of a specific antibody and was reproducible (coefficient of variation, CV 3.6%).

Provider-patient communication and metabolic control

OBJECTIVE

To determine whether nurses' and NIDDM patients' communication styles during consultations are related to subsequent metabolic control and to examine factors influencing patterns of communication in these consultations.

RESEARCH DESIGN AND METHODS

A total of 47 NIDDM patients participated in the study and completed the following procedures: 1) assessment of baseline HbA1, 2) attended 3.5 days of diabetes education, 3) returned in 1 mo for a follow-up consultation with a nurse, and (4) returned in 9-12 wk for a follow-up HbA1 assessment. The communication variables coded from the consultations were the frequency with which nurses produced controlling, informative, and patient-centered utterances and the frequency with which patients sought information, engaged in decision making, and expressed negative affect.

RESULTS

The results were as follows: 1) patients experienced poorer metabolic control after interacting with nurses who were more controlling and directive in their communication with patients (r = 0.39, P < 0.01); 2) the nurses' use of patient-centered responses was directly related to the degree to which patients expressed feelings (r = 0.34, P < 0.01) and exhibited decision-making behavior (r = 0.62, P < 0.01); and 3) several of the nurses' and patients' communicative behaviors were related to patient characteristics such as age, sex, education, and baseline HbA1 levels.

CONCLUSIONS

The results suggest that providers' attempts to exert considerable control during consultations with NIDDM patients may be counterproductive and contribute to poorer outcomes. The findings also indicate that patient-centered behaviors (e.g., encouraging the patient's involvement, respecting the patient's opinion, and offering support) facilitate the patient's ability to be an active participant in the consultation.

A kinetic study of the photochemical inactivation of adenylate kinases of Mycobacterium marinum and bovine heart mitochondria

Using incident light energy of about 76 mW.cm-2 in a dye-sensitized photooxidation reaction, we have investigated the possible involvement of one or both of the histidine residues in the catalytic activity of adenylate kinase (ATP:AMP phosphotransferase) of Mycobacterium marinum. We have done this by investigating the kinetics of photochemical inactivation of the enzyme. At pH 7.4, the kinetics of photoinactivation are biphasic with two different pseudo-first-order rate constants. Adenosine 5'-pentaphospho 5'-adenosine (Ap5A), ATP and, to some extent, AMP, all gave protection to the enzyme from inactivation. Amino-acid analysis of the photoinactivated enzyme indicated the loss of the two histidine residues. This, and the fact that photoinactivation occurred faster at alkaline compared to acidic pH, indicated the involvement of the histidine residues in the catalytic activity. A mathematical model is developed which assumes that both histidine residues are required for maximal catalytic activity: one is located peripherally, is exposed, and therefore is readily photooxidized (pseudo-first-order rate constant, k1 = 1.3.10(-2)s-1), while the other is located at the active site, involved in substrate-binding and is shielded (pseudo-first-order rate constant, k2 = 2.9.10(-4)s-1). However, this shielded histidine could be exposed and made more accessible to photooxidation either by raising the pH above 10, or alternatively, by the addition of 8 M acetamide (or 6 M guanidine). Under these conditions, which apparently cause unfolding of the protein molecule, the kinetics of photoinactivation change from biphasic to monophasic, suggesting that both histidine residues are equally exposed and are photooxidized at the same rate. Unlike the enzyme from M. marinum, adenylate kinase from bovine heart mitochondria shows monophasic kinetics of photoinactivation at pH 7.4, suggesting that only one of the six histidine residues is essential for catalytic activity, or if more than one, then they all must be equally exposed. Further, ATP, AMP or Ap5A did not provide protection against photoinactivation, suggesting that the histidine residue(s) involved in the catalytic activity must remain exposed after the substrates bind at the active site of the mitochondrial enzyme.

Resource management in a distributed internetwork environment

The resource management system is designed to support location-transparent access to resources and to improve performance in a distributed internetwork environment. In this environment, access to one of several machines that have a given resource is determined by the effective bandwidth and reliability of internetwork communications, the ability to interact with the machines offering the resources, and the load on those machines. Using these criteria, the resources management system enables applications to determine and obtain access to the “best” copy of a resource. Should the resource become unavailable from one machine, the resource management system can dynamically direct the applications to another machine known to offer the same resource. This paper discusses some of the problems that motivated the design, an implementation for a UNIX workstation environment, and future directions for the resource management system .