Data-driven hypothesis generation among inexperienced clinical researchers: A comparison of secondary data analyses with visualization (VIADS) and other tools

Objectives

To compare how clinical researchers generate data-driven hypotheses with a visual interactive analytic tool (VIADS, a visual interactive analysis tool for filtering and summarizing large datasets coded with hierarchical terminologies) or other tools.

Methods

We recruited clinical researchers and separated them into "experienced" and "inexperienced" groups. Participants were randomly assigned to a VIADS or control group within the groups. Each participant conducted a remote 2-hour study session for hypothesis generation with the same study facilitator on the same datasets by following a think-aloud protocol. Screen activities and audio were recorded, transcribed, coded, and analyzed. Hypotheses were evaluated by seven experts on their validity, significance, and feasibility. We conducted multilevel random effect modeling for statistical tests.

Results

Eighteen participants generated 227 hypotheses, of which 147 (65%) were valid. The VIADS and control groups generated a similar number of hypotheses. The VIADS group took a significantly shorter time to generate one hypothesis (e.g., among inexperienced clinical researchers, 258 s versus 379 s, p = 0.046, power = 0.437, ICC = 0.15). The VIADS group received significantly lower ratings than the control group on feasibility and the combination rating of validity, significance, and feasibility.

Conclusion

The role of VIADS in hypothesis generation seems inconclusive. The VIADS group took a significantly shorter time to generate each hypothesis. However, the combined validity, significance, and feasibility ratings of their hypotheses were significantly lower. Further characterization of hypotheses, including specifics on how they might be improved, could guide future tool development.

Data-driven hypothesis generation among inexperienced clinical researchers: A comparison of secondary data analyses with visualization (VIADS) and other tools

Objectives

To compare how clinical researchers generate data-driven hypotheses with a visual interactive analytic tool (VIADS, a visual interactive analysis tool for filtering and summarizing large data sets coded with hierarchical terminologies) or other tools.

Methods

We recruited clinical researchers and separated them into "experienced" and "inexperienced" groups. Participants were randomly assigned to a VIADS or control group within the groups. Each participant conducted a remote 2-hour study session for hypothesis generation with the same study facilitator on the same datasets by following a think-aloud protocol. Screen activities and audio were recorded, transcribed, coded, and analyzed. Hypotheses were evaluated by seven experts on their validity, significance, and feasibility. We conducted multilevel random effect modeling for statistical tests.

Results

Eighteen participants generated 227 hypotheses, of which 147 (65%) were valid. The VIADS and control groups generated a similar number of hypotheses. The VIADS group took a significantly shorter time to generate one hypothesis (e.g., among inexperienced clinical researchers, 258 seconds versus 379 seconds, p = 0.046, power = 0.437, ICC = 0.15). The VIADS group received significantly lower ratings than the control group on feasibility and the combination rating of validity, significance, and feasibility.

Conclusion

The role of VIADS in hypothesis generation seems inconclusive. The VIADS group took a significantly shorter time to generate each hypothesis. However, the combined validity, significance, and feasibility ratings of their hypotheses were significantly lower. Further characterization of hypotheses, including specifics on how they might be improved, could guide future tool development.

Development, validation, and usage of metrics to evaluate the quality of clinical research hypotheses

Objectives

Metrics and instruments can provide guidance for clinical researchers to assess their potential research projects at an early stage before significant investment. Furthermore, metrics can also provide structured criteria for peer reviewers to assess others' clinical research manuscripts or grant proposals. This study aimed to develop, test, validate, and use evaluation metrics and instruments to accurately, consistently, and conveniently assess the quality of scientific hypotheses for clinical research projects.

Materials and Methods

Metrics development went through iterative stages, including literature review, metrics and instrument development, internal and external testing and validation, and continuous revisions in each stage based on feedback. Furthermore, two experiments were conducted to determine brief and comprehensive versions of the instrument.

Results

The brief version of the instrument contained three dimensions: validity, significance, and feasibility. The comprehensive version of metrics included novelty, clinical relevance, potential benefits and risks, ethicality, testability, clarity, interestingness, and the three dimensions of the brief version. Each evaluation dimension included 2 to 5 subitems to evaluate the specific aspects of each dimension. For example, validity included clinical validity and scientific validity. The brief and comprehensive versions of the instruments included 12 and 39 subitems, respectively. Each subitem used a 5-point Likert scale.

Conclusion

The validated brief and comprehensive versions of metrics can provide standardized, consistent, and generic measurements for clinical research hypotheses, allow clinical researchers to prioritize their research ideas systematically, objectively, and consistently, and can be used as a tool for quality assessment during the peer review process.

Keyphrase Identification Using Minimal Labeled Data with Hierarchical Context and Transfer Learning

Interoperable clinical decision support system (CDSS) rules provide a pathway to interoperability, a well-recognized challenge in health information technology. Building an ontology facilitates creating interoperable CDSS rules, which can be achieved by identifying the keyphrases (KP) from the existing literature. However, KP identification for data labeling requires human expertise, consensus, and contextual understanding. This paper aims to present a semi-supervised KP identification framework using minimal labeled data based on hierarchical attention over the documents and domain adaptation. Our method outperforms the prior neural architectures by learning through synthetic labels for initial training, document-level contextual learning, language modeling, and fine-tuning with limited gold standard label data. To the best of our knowledge, this is the first functional framework for the CDSS sub-domain to identify KPs, which is trained on limited labeled data. It contributes to the general natural language processing (NLP) architectures in areas such as clinical NLP, where manual data labeling is challenging, and light-weighted deep learning models play a role in real-time KP identification as a complementary approach to human experts' effort.

Comparative performances of nomograms and conditional survival after resection of adrenocortical cancer

Background

Adrenocortical carcinomas (ACCs) carry a poor prognosis. This study assessed the comparative performance of existing nomograms in estimating the likelihood of survival, along with the value of conditional survival estimation for patients who had already survived for a given length of time after surgery.

Methods

This was an observational study based on a prospectively developed departmental database that recorded details of patients operated for ACC in a UK tertiary referral centre.

Results

Of 74 patients with ACC managed between 2001 and 2020, data were analysed for 62 patients (32 women and 30 men, mean(s.d.) age 51(17) years) who had primary surgical treatment in this unit. Laparoscopic (9) or open adrenalectomies (53) were performed alone or in association with a multivisceral resection (27). Most of the tumours were left-sided (40) and 18 were cortisol-secreting.

Overall median survival was 33 months, with 1-, 3- and 5-year survival rates of 79, 49, and 41 per cent respectively. Age over 55 years, higher European Network for Study of Adrenal Tumours stage, and cortisol secretion were associated with poorer survival in univariable analyses. Four published nomograms suggested widely variable outcomes that did not correlate with observed overall survival at 1, 3 or 5 years after operation. The 3-year conditional survival at 2 years (probability of surviving to postoperative year 5) was 65 per cent, compared with a 5-year actuarial survival rate of 41 per cent calculated from the time of surgery.

Conclusion

Survival of patients with ACC correlates with clinical parameters but not with published nomograms. Conditional survival might provide a more accurate estimate of survival for patients who have already survived for a certain amount of time after resection.

Neoadjuvant radiotherapy in rectal cancer - less is more?

AIM

There is significant international variation in the use of neoadjuvant radiation prior to total mesorectal excision. The MERCURY group advocate selective neoadjuvant chemoradiotherapy (CRT). We have performed a retrospective, single-centre study of patients treated with CRT, where only the circumferential resection margin is threatened, with the aim of identifying whether a more selective approach to CRT provides acceptable local relapse rates (LRRs).

METHOD

All consecutive patients who underwent radical surgery for rectal adenocarcinoma over a 5-year period (2007-2012) in the Oxford University Trust were considered. Electronic hospital systems were reviewed to obtain patient and tumour demographics, treatment and follow-up information. All patients were classified into risk categories according to National Institute for Health and Care Excellence guidance. Data were analysed using Microsoft Excel and R.

RESULTS

Two hundred and seventy-two patients were identified: 123, 89 and 60 in the high-, intermediate- and low-risk categories, respectively. Seventy-nine per cent of those in the high-risk group, 6% in the intermediate and 5% in the low-risk group underwent CRT. The overall 5-year LRR and distant recurrence rate (DRR) were 5.2% and 17.8%, respectively. The 5-year LRR for those who went straight to surgery was 2.0% and for those who had neoadjuvant CRT it was 7.4%. The DRR for these two groups was 8.5% and 18.9%, respectively.

CONCLUSION

Our series demonstrates that the use of CRT only in margin-threatening tumours, results in an exceptionally low LRR for those without margin-threatening disease. In routine clinical care, this strategy can minimize the significant morbidity of multimodal treatment and allow earlier introduction of systemic therapy to minimize distant recurrence.

Substrate Supporting Disc Method for Confirmed Detection of Total Coliforms and E. coli in all Foods: Collaborative Study

The Coli Complete® substrate supporting disc (SSD) method for simultaneous confirmed total coliform count and Escherichia coli determination in all foods was compared with AOAC most probable number (MPN) methods, 966.23 and 966.24. Twenty-nine laboratories participated in this collaborative study in which 6 food types were analyzed. Four food types, raw ground beef, pork sausage, raw liquid milk, and nut meats, were naturally contaminated with coliform bacteria. Two foods, dry egg and fresh frozen vegetables, were seeded with coliforms. Three food types, ground beef, raw liquid milk, and pork sausage, were naturally contaminated with E. coli. Although pork sausage was naturally contaminated, the level was very low (<10/50 g); therefore, additional E. coli were inoculated into 1 lot of this food type. Three food types, nut meats, dry egg, and fresh frozen vegetables, were inoculated with E. coli. For naturally contaminated samples, duplicate determinations were made on 3 separate lots for each food type. For inoculated samples, low, medium, and high contamination levels plus uninoculated control samples were examined in duplicate. Data were analyzed separately for total coliform bacteria and for E. coli. Mean log MPN counts were determined by the SSD method and the appropriate AOAC MPN method. Results were then analyzed for repeatability, reproducibility, and mean log MPN statistical equivalence. Results were statistically equivalent for all total coliform levels in all food types except frozen vegetable and raw nut meat uninoculated control samples and 1 lot of pork sausage where the SSD method produced statistically significant greater numbers. For the E. coli determinations, results were statistically equivalent across all samples and all levels for each food type. The SSD method has been adopted first action by AOAC International for confirmed detection of total coliforms and E. coli in all foods.

Mortality of civilian patients with suspected traumatic haemorrhage receiving pre-hospital transfusion of packed red blood cells compared to pre-hospital crystalloid

Background

Major haemorrhage is a leading cause of mortality following major trauma. Increasingly, Helicopter Emergency Medical Services (HEMS) in the United Kingdom provide pre-hospital transfusion with blood products, although the evidence to support this is equivocal. This study compares mortality for patients with suspected traumatic haemorrhage transfused with pre-hospital packed red blood cells (PRBC) compared to crystalloid.

Methods

A single centre retrospective observational cohort study between 1 January 2010 and 1 February 2015. Patients triggering a pre-hospital Code Red activation were eligible. The primary outcome measure was all-cause mortality at 6 hours (h) and 28 days (d), including a sub-analysis of patients receiving a major and massive transfusion. Multivariable regression models predicted mortality. Multiple Imputation was employed, and logistic regression models were constructed for all imputed datasets.

Results

The crystalloid (n = 103) and PRBC (n = 92) group were comparable for demographics, Injury Severity Score (p = 0.67) and mechanism of injury (p = 0.73). Observed 6 h mortality was smaller in the PRBC group (n = 10, 10%) compared to crystalloid group (n = 19, 18%). Adjusted OR was not statistically significant (OR 0.48, CI 0.19–1.19, p = 0.11). Observed mortality at 28 days was smaller in the PRBC group (n = 21, 26%) compared to crystalloid group (n = 31, 40%), p = 0.09. Adjusted OR was not statistically significant (OR 0.66, CI 0.32–1.35, p = 0.26). A statistically significant greater proportion of the crystalloid group required a major transfusion (n = 62, 60%) compared to the PRBC group (n = 41, 40%), p = 0.02. For patients requiring a massive transfusion observed mortality was smaller in the PRBC group at 28 days (p = 0.07).

Conclusion

In a single centre UK HEMS study, in patients with suspected traumatic haemorrhage who received a PRBC transfusion there was an observed, but non-significant, reduction in mortality at 6 h and 28 days, also reflected in a massive transfusion subgroup. Patients receiving pre-hospital PRBC were significantly less likely to require an in-hospital major transfusion. Further adequately powered multi-centre prospective research is required to establish the optimum strategy for pre-hospital volume replacement in patients with traumatic haemorrhage.

The impact of providing blood to the scene of an accident on transfusion laboratory practice

BACKGROUND

Haemorrhage is the leading cause of mortality during trauma. In 2012, London's Air Ambulance introduced Blood on Board (BOB), transfusing group O red cells (RBC) to trauma patients at the scene.

OBJECTIVES

This study assessed the impact of BOB on the number of mixed field samples received by the laboratory, the number of group O RBC transfused to non-group O patients and the ratio of RBC to fresh frozen plasma (FFP) transfused in the initial 24 h.

METHODS

Three major trauma centres collected data on patients for whom the major haemorrhage protocol was activated between August 2008 and February 2012 pre-BOB and March 2012 and December 2013 post-BOB.

RESULTS

A total of 233 trauma patients were identified pre-BOB and 119 post-BOB. There was no significant difference in the percentage of group O units transfused to non-group O patients (75 vs 82%, P = 0·21) or the RBC : FFP ratio (pre-BOB mean 1·6 [interquartile range (IQR) 1·0-2·0]; post-BOB mean 1·7 [IQR 1·1-2·2], P = 0·24). There was no significant difference in the percentage of mixed field samples received (23% vs 27%, P = 0·3).

CONCLUSION

The introduction of BOB did not change the proportion of group O RBC transfused or the RBC : FFP ratio; however, the proportion of acceptable samples decreased. This is largely due to an increase in blood samples not received from the post-BOB cohort, which we believe is probably due to patients who died at the scene. We have introduced robust systems to indicate reasons for not obtaining samples.

Ocean Data Assimilation in Support of Climate Applications: Status and Perspectives

Ocean data assimilation brings together observations with known dynamics encapsulated in a circulation model to describe the time-varying ocean circulation. Its applications are manifold, ranging from marine and ecosystem forecasting to climate prediction and studies of the carbon cycle. Here, we address only climate applications, which range from improving our understanding of ocean circulation to estimating initial or boundary conditions and model parameters for ocean and climate forecasts. Because of differences in underlying methodologies, data assimilation products must be used judiciously and selected according to the specific purpose, as not all related inferences would be equally reliable. Further advances are expected from improved models and methods for estimating and representing error information in data assimilation systems. Ultimately, data assimilation into coupled climate system components is needed to support ocean and climate services. However, maintaining the infrastructure and expertise for sustained data assimilation remains challenging.

Negative pressure wound therapy with Bio-Dome dressing technology in the treatment of complex wounds: a case series

The treatment of complex wounds is difficult and not always effective. Various treatment options are used with varying degrees of success. Negative pressure wound therapy (NPWT) is a cost-efficient and effective way to help treat these wounds. The use of a vacuum device applies the negative pressure to the site of the wound and promotes waste removal and increases circulation and tissue formation. While various NPWT systems are currently on the market, we utilised the ConvaTec Engenex® system with Bio-DomeTM technology; however, our case study is not intended to advocate the specific use of this system, but instead focuses on the use of NPWT as a viable option for wound healing. Each of the following case study patients presented with difficult-to-heal wounds that failed traditional therapeutic approaches. Through the use of NPWT, our patients saw major wound size reductions. Each patient exhibited at least a 94% reduction in wound area, wound volume or both.

Selective case study describing the use of Apligraf on necrobiosis lipoidica associated with diabetes

Necrobiosis lipoidica is a rare skin disease characterised by large, well-demarcated, symmetrical plaques with overlying telangiectasias and atrophic, fibrotic features. The disease is associated with diabetes mellitus (1 in 300 cases), but can also be linked to other diseases such as rheumatoid arthritis. Women are three times more likely to develop necrobiosis lipoidica compared to men. Ulcerations are the most serious type of complications in necrobiosis lipoidica, and they occur most frequently on the legs of patients. However, the aetiology of necrobiosis lipoidica still remains unclear. Although many studies have been conducted in order to determine necrobiosis lipoidica's pathophysiology, a clear and definite path to disease has not been recorded. In this case study, a patient with necrobiosis lipoidica that had been refractory to conventional therapy received treatment with Apligraf® bioengineered wound dressings. Apligraf was shown to be effective in managing the patient's multiple hard-to-heal wounds. It was more successful than previous therapies in achieving granulation tissue formation and wound volume reduction, in addition to being a more rapid form of treatment.