A machine-learning algorithm for distinguishing malignant from benign indeterminate thyroid nodules using ultrasound radiomic features

Background: Ultrasound (US)-guided fine needle aspiration (FNA) cytology is the gold standard for the evaluation of thyroid nodules. However, up to 30% of FNA results are indeterminate, requiring further testing. In this study, we present a machine-learning analysis of indeterminate thyroid nodules on ultrasound with the aim to improve cancer diagnosis. Methods: Ultrasound images were collected from two institutions and labeled according to their FNA (F) and surgical pathology (S) diagnoses [malignant (M), benign (B), and indeterminate (I)]. Subgroup breakdown (FS) included: 90 BB, 83 IB, 70 MM, and 59 IM thyroid nodules. Margins of thyroid nodules were manually annotated, and computerized radiomic texture analysis was conducted within tumor contours. Initial investigation was conducted using five-fold cross-validation paradigm with a two-class Bayesian artificial neural networks classifier, including stepwise feature selection. Testing was conducted on an independent set and compared with a commercial molecular testing platform. Performance was evaluated using receiver operating characteristic analysis in the task of distinguishing between malignant and benign nodules. Results: About 1052 ultrasound images from 302 thyroid nodules were used for radiomic feature extraction and analysis. On the training/validation set comprising 263 nodules, five-fold cross-validation yielded area under curves (AUCs) of 0.75 [Standard Error (SE) = 0.04; P < 0.001 ] and 0.67 (SE = 0.05; P = 0.0012 ) for the classification tasks of MM versus BB, and IM versus IB, respectively. On an independent test set of 19 IM/IB cases, the algorithm for distinguishing indeterminate nodules yielded an AUC value of 0.88 (SE = 0.09; P < 0.001 ), which was higher than the AUC of a commercially available molecular testing platform (AUC = 0.81, SE = 0.11; P < 0.005 ). Conclusion: Machine learning of computer-extracted texture features on gray-scale ultrasound images showed promising results classifying indeterminate thyroid nodules according to their surgical pathology.

A Survey of American Thyroid Association Members Regarding the 2015 Adult Thyroid Nodule and Differentiated Thyroid Cancer Clinical Practice Guidelines

Background: The 2015 American Thyroid Association (ATA) clinical practice guidelines (CPGs) on management of thyroid nodules (TNs) and differentiated thyroid cancer (DTC) in adults were developed to inform clinicians, patients, researchers, and health policy makers about the best available evidence, and its limitations, relating to management of these conditions. Methods: We conducted a cross-sectional electronic survey of ATA members' perspectives of these CPGs, using a standardized survey (Clinician Guidelines Determinant Questionnaire) developed by the Guidelines International Network. A survey link was electronically mailed to members in February of 2019, with reminders sent to nonrespondents 2 and 5 weeks later. Data were descriptively summarized, after excluding missing responses. Results: The overall response rate was 19.8% (348/1761). The effective response rate was 20.2% (348/1720), after excluding a recently deceased member and individuals who had either invalid e-mail addresses or whose e-mails were returned. Of the respondents, 37.9% (132/348) were female, 60.4% (209/346) were endocrinologists, 27.5% (95/346) were surgeons, and 3.5% (12/346) were nuclear medicine specialists. The majority of respondents (71.9%; 250/348) were at a mid- or advanced-career level, and more than half were in academia (57.5%; 195/339). The majority (69.8%; 243/348) practiced in North America. The vast majority of respondents indicated that the CPGs explained the underlying evidence (92.3%; 298/323) and 92.9% (300/323) agreed or strongly agreed with the content. Most respondents stated that they regularly used the CPGs in their practice (83.0%; 268/323). Most respondents (83.0%; 268/323) also agreed or strongly agreed that the recommendations were easy to incorporate in their practice. The most popular CPG format was an electronic desktop file (78.8%; 252/320). Shorter more frequent CPGs were favored by 55.0% (176/320) of respondents, and longer traditional CPGs were favored by 39.7% (127/320). Conclusions: The clinical content and evidence explanations in the adult TN and DTC CPGs are widely accepted and applied among ATA survey respondents. Future ATA CPG updates need to be optimized to best meet users' preferences regarding format, frequency, and length.

Distributed Matching with Mixed Maximum-Minimum Utilities

In this article, we study distributed agent matching with search friction in environments characterized by costly exploration, where each agent’s utility from forming a partnership is influenced by some linear combination of the maximum and the minimum among the two agents’ competence. The article provides a cohesive analysis for such case, proving the equilibrium structure for the different min-max linear combinations that may be used. The article presents an extensive equilibrium analysis of such settings, proving three distinct resulting patterns of the acceptance thresholds used by the different agents. The first relates to settings where a greater emphasis is placed on the minimum type, or in the extreme case where the minimum type solely determines the output. In these cases, the assortative matching characteristic holds, where all agents set their threshold below their own type and the greater is the agent type, the greater is its threshold. When the utility from the partnership formation is solely determined by the maximum type, we show that there exists a type x * where partnerships form if and only if one of the agents has a type equal to or greater than x * . When a greater emphasis is placed on the maximum type (but not only), we prove that assortative matching never holds, and the change in the agents’ acceptance thresholds can frequently shift from an increase to a decrease.

Constraining Information Sharing to Improve Cooperative Information Gathering

This paper considers the problem of cooperation between self-interested agents in acquiring better information regarding the nature of the different options and opportunities available to them. By sharing individual findings with others, the agents can potentially achieve a substantial improvement in overall and individual expected benefits. Unfortunately, it is well known that with self-interested agents equilibrium considerations often dictate solutions that are far from the fully cooperative ones, hence the agents do not manage to fully exploit the potential benefits encapsulated in such cooperation. In this paper we introduce, analyze and demonstrate the benefit of five methods aiming to improve cooperative information gathering. Common to all five that they constrain and limit the information sharing process. Nevertheless, the decrease in benefit due to the limited sharing is outweighed by the resulting substantial improvement in the equilibrium individual information gathering strategies. The equilibrium analysis given in the paper, which, in itself is an important contribution to the study of cooperation between self-interested agents, enables demonstrating that for a wide range of settings an improved individual expected benefit is achieved for all agents when applying each of the five methods.

Competitive Shopbots-Mediated Markets

This article considers markets mediated by autonomous self-interested comparison-shopping agents. As in today’s markets, the agents do not charge buyers for their services but rather benefit from payments obtained from sellers upon the execution of a transaction. The agents aim at maximizing their expected benefit, taking into consideration the cost incurred by the search and competition dynamics that arise in the multi-agent setting. This article provides a comprehensive analysis of such models, based on search theory principles. The analysis results in a characterization of the buyers’ and agents’ search strategies in equilibrium. The main result of this article is that the use of self-interested comparison-shopping agents can result in a beneficial equilibrium, where both buyers and sellers benefit, in comparison to the case where buyers control the comparison-shopping agent, and the comparison-shopping agents necessarily do not lose. This, despite the fact that the service is offered for free to buyers and its cost is essentially covered by sellers. The analysis generalizes to any setting where buyers can use self-interested agents capable of effectively performing the search (e.g., evaluating opportunities) on their behalf.

A trial of inpatient indication based prescribing during computerized order entry with medications commonly used off-label

BACKGROUND

Requiring indications for inpatient medication orders may improve the quality of prescribing and allow for easier placement of diagnoses on the problem list. Indications for inpatient medication orders are also required by some regulators.

OBJECTIVE

This study assessed a clinical decision support (CDS) system designed to obtain indications and document problems during inpatient computerized physician order entry (CPOE) of medications frequently used off-label.

METHODS

A convenience sample of three medications frequently used off-label were selected: the PPI lansoprazole; intravenous immune globulin, and recombinant Factor VIIa. Alerts triggered when a medication was ordered without an FDA approved indication in the problem list. The alerts prompted clinicians to enter either a labeled or off-label indication for the order. Chart review was used as the gold standard to assess the accuracy of clinician entered information.

RESULTS

The PPI intervention generated 873 alerts during 60 days of operation; IVIG 55 alerts during alerts during 93 days; Factor VIIa 25 alerts during 175 days. Agreement between indications entered and chart review was 63% for PPI, 49% for IVIG, and 29% for Factor VIIa. The alerts for PPI, IVIG and Factor VIIa alerts produced accurate diagnoses for the problem list 9%, 16% and 24% respectively. Rates of off-label use measured by chart review were 87% for PPI, and 100% for IVIG and factor VIIa, which were higher than if measured using the ordering clinicians' indications.

CONCLUSION

This trial of indication-based prescribing using CDS and CPOE produced less than optimal accuracy of the indication data as well as a low yield of accurate problems placed on the problem list. These results demonstrate the challenge inherent in obtaining accurate indication information during prescribing and should raise concerns over potential mandates for indication based prescribing and motivate further study of appropriate mechanisms to obtain indications during CPOE.

Cooperative Search with Concurrent Interactions

In this paper we show how taking advantage of autonomous agents' capability to maintain parallel interactions with others, and incorporating it into the cooperative economic search model results in a new search strategy which outperforms current strategies in use. As a framework for our analysis we use the electronic marketplace, where buyer agents have the incentive to search cooperatively. The new search technique is quite intuitive, however its analysis and the process of extracting the optimal search strategy are associated with several significant complexities. These difficulties are derived mainly from the unbounded search space and simultaneous dual affects of decisions taken along the search. We provide a comprehensive analysis of the model, highlighting, demonstrating and proving important characteristics of the optimal search strategy. Consequently, we manage to come up with an efficient modular algorithm for extracting the optimal cooperative search strategy for any given environment. A computational based comparative illustration of the system performance using the new search technique versus the traditional methods is given, emphasizing the main differences in the optimal strategy's structure and the advantage of using the proposed model.

External radiation and thyroid neoplasia

Radiation remains the only factor that has been shown unequivocally to cause (nonmedullary) thyroid cancer. Recent advances include the analysis of the dose-response relationship using data pooled from multiple studies. This analysis confirms that radiation-induced thyroid cancers continue to occur, with a maximum risk at approximately 30 years after exposure. Physicians asked to evaluate patients with a history of radiation exposure should attempt to estimate the dose from the history and should be familiar with the other risk factors. For some individuals, screening should include thyroid imaging, but the results of such imaging, especially with thyroid ultrasound, should be interpreted with caution. The treatment of radiation-induced thyroid cancers is based on the observation that they appear to be no more aggressive than thyroid cancers not associated with radiation. In the future, more information should emerge about the role of cancer genes and susceptibility factors in radiation-induced thyroid cancer.

Resistance to thyroid hormone in subjects from two unrelated families is associated with a point mutation in the thyroid hormone receptor beta gene resulting in the replacement of the normal proline 453 with serine

Resistance to thyroid hormone (RTH) is a condition of impaired tissue responsiveness to thyroid hormone characterized by elevated free thyroid hormone levels in serum accompanied by nonsuppressed TSH. RTH has been associated with mutations in the thyroid hormone receptor (TR) beta gene. We report studies carried out in 9 members of a family (F94) of Jewish ethnic origin and a single subject of Mexican origin. All subjects fulfilling the criteria of RTH (6 of family F94 and one of family F27) had the same point mutation in the T3-binding domain on one of the two alleles of the TR beta gene. This mutation resulted in the replacement of the normal proline-453 with serine (P453S). Nevertheless, the clinical characteristics of affected members of each of the two families differed as did the severity of hormonal resistance in terms of responses to the administration of L-T3. Genetic studies indicate that the same mutation occurred independently in each of the two families.

Sequencing of the variant thyroxine-binding globulin (TBG)-San Diego reveals two nucleotide substitutions

Thyroxine-binding globulin (TBG) is a liver glycoprotein that transports thyroid hormone in serum. In 1989, a variant TBG was reported with reduced binding affinity for thyroxine (T4) and triiodothyronine (T3) which results in low serum T4 and T3 levels. This variant, TBG-San Diego (TBG-SD), also displays reduced heat stability but has a normal isoelectric focusing pattern. We now report the sequence of the entire coding region of TBG-San Diego. It reveals two nucleotide substitutions: one located in exon 1 which results in the replacement of the normal Ser-23 (TCA) with threonine (ACA) and the other, located in exon 3, changes the normal codon 283 of TTG (leucine) with that of TTT, (phenylalanine). Allele specific amplification was used to search for both nucleotide substitutions in four affected members of the family. Results confirmed the co-segregation of these nucleotide substitutions with the TBG-SD phenotype. The substitution in codon 283 has been previously described and exists as a polymorphism in some ethnic groups or in combination with other TBG variants with different physical characteristics. Thus, it appears that the replacement of Ser-23 with threonine is responsible for the observed alterations in physical properties of TBG-San Diego.

Elevated serum thyroglobulin level in congenital thyroxine-binding globulin deficiency

Serum thyroglobulin (TG) is normally under TSH control. Serum TG levels are elevated during increased thyroid gland stimulation and suppressed by exogenous thyroid hormone. High serum TG levels are also found with thyroid gland damage and in patients with differentiated thyroid neoplasms. Congenital T4-binding globulin deficiency was found in this study to be an additional condition in which serum TG levels may be elevated. Elevated TG levels were found in 13 of 36 patients (36%) with congenital TBG deficiency compared to 1 of 27 unaffected relatives (4%). Mean TSH and free T4 index values were not significantly different. A postulated mechanism involves transient TSH stimulation of the thyroid after transient small declines in circulating free hormone levels due to the decreased extrathyroidal pool of thyroid hormone associated with T4-binding globulin deficiency.