Search

Interest in the application of machine learning (ML) techniques to medicine is growing fast and wide because of their ability to endow decision support systems with so-called artificial intelligence, particularly in those medical disciplines that extensively rely on digital imaging. Nonetheless, achieving a pragmatic and ecological validation of medical AI systems in real-world settings is difficult, even when these systems exhibit very high accuracy in laboratory settings. This difficulty has been called the "last mile of implementation." In this review of the concept, we claim that this metaphorical mile presents two chasms: the hiatus of human trust and the hiatus of machine experience. The former hiatus encompasses all that can hinder the concrete use of AI at the point of care, including availability and usability issues, but also the contradictory phenomena of cognitive ergonomics, such as automation bias (overreliance on technology) and prejudice against the machine (clearly the opposite). The latter hiatus, on the other hand, relates to the production and availability of a sufficient amount of reliable and accurate clinical data that is suitable to be the "experience" with which a machine can be trained. In briefly reviewing the existing literature, we focus on this latter hiatus of the last mile, as it has been largely neglected by both ML developers and doctors. In doing so, we argue that efforts to cross this chasm require data governance practices and a focus on data work, including the practices of data awareness and data hygiene. To address the challenge of bridging the chasms in the last mile of medical AI implementation, we discuss the six main socio-technical challenges that must be overcome in order to build robust bridges and deploy potentially effective AI in real-world clinical settings., Competing Interests: Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at http://dx.doi.org/10.21037/atm.2020.03.63). FC serves as an unpaid editorial board member of Annals of Translational Medicine from Apr 2018 to Mar 2020. The other authors have no conflicts of interest to declare., (2020 Annals of Translational Medicine. All rights reserved.)

Background and Objectives: Collecting Patient-Reported Outcomes (PROs) is an important way to get first-hand information by patients on the outcome of treatments and surgical procedure they have undergone, and hence about the quality of the care provided. However, the quality of PRO data cannot be given for granted and cannot be traced back to the dimensions of timeliness and completeness only. While the reliability of these data can be guaranteed by adopting standard and validated questionnaires that are used across different health care facilities all over the world, these facilities must take responsibility to assess, monitor and ensure the validity of PROs that are collected from their patients. Validity is affected by biases that are hidden in the data collected. This contribution is then aimed at measuring bias in PRO data, for the impact that these data can have on clinical research and post-marketing surveillance., Methods: We considered the main biases that can affect PRO validity: Response bias, in terms of Acquiescence bias and Fatigue bias; and Non-Response bias. To assess Acquiescence bias, phone interviews and online surveys were compared, adjusted by age. To assess Fatigue bias, we proposed a specific item about session length and compared PROs scores stratifying according to the responses to this item. We also calculated the intra-patient agreement by conceiving an intra-interview test-retest. To assess Non-Response bias, we considered patients who participated after the saturation of the response-rate curve as proxy of potential non respondents and compared the outcomes in these two strata. All methods encompassed common statistical techniques and are cost-effective at any facility collecting PRO data., Results: Acquiescence bias resulted in significantly different scores between patients reached by either phone or email. In regard to Fatigue bias, stratification by perceived fatigue resulted in contrasting results. A relevant difference was found in intra-patient agreement and an increasing difference in average scores as a function of interview length (or completion time). In regard to Non-Response bias, we found non-significant differences both in scores and variance., Conclusions: In this paper, we present a set of cost-effective techniques to assess the validity of retrospective PROs data and share some lessons learnt from their application at a large teaching hospital specialized in musculoskeletal disorders that collects PRO data in the follow-up phase of surgery performed therein. The main finding suggests that response bias can affect the PRO validity. Further research on the effectiveness of simple and cost-effective solutions is necessary to mitigate these biases and improve the validity of PRO data., (Copyright © 2019 Elsevier B.V. All rights reserved.)

This article focuses on the production side of clinical data work, or data recording work, and in particular, on its multiplicity in terms of data variability. We report the findings from two case studies aimed at assessing the multiplicity that can be observed when the same medical phenomenon is recorded by multiple competent experts, yet the recorded data enable the knowledgeable management of illness trajectories. Often framed in terms of the latent unreliability of medical data, and then treated as a problem to solve, we argue that practitioners in the health informatics field must gain a greater awareness of the natural variability of data inscribing work, assess it, and design solutions that allow actors on both sides of clinical data work, that is, the production and care, as well as the primary and secondary uses of data to aptly inform each other's practices.

Problem: Clinical practice requires the production of a time- and resource-consuming great amount of notes. They contain relevant information, but their secondary use is almost impossible, due to their unstructured nature. Researchers are trying to address this problems, with traditional and promising novel techniques. Application in real hospital settings seems not to be possible yet, though, both because of relatively small and dirty dataset, and for the lack of language-specific pre-trained models. Aim: Our aim is to demonstrate the potential of the above techniques, but also raise awareness of the still open challenges that the scientific communities of IT and medical practitioners must jointly address to realize the full potential of unstructured content that is daily produced and digitized in hospital settings, both to improve its data quality and leverage the insights from data-driven predictive models. Methods: To this extent, we present a narrative literature review of the most recent and relevant contributions to leverage the application of Natural Language Processing techniques to the free-text content electronic patient records. In particular, we focused on four selected application domains, namely: data quality, information extraction, sentiment analysis and predictive models, and automated patient cohort selection. Then, we will present a few empirical studies that we undertook at a major teaching hospital specializing in musculoskeletal diseases. Results: We provide the reader with some simple and affordable pipelines, which demonstrate the feasibility of reaching literature performance levels with a single institution non-English dataset. In such a way, we bridged literature and real world needs, performing a step further toward the revival of notes fields.

In this paper we present the findings of a systematic literature review covering the articles published in the last two decades in which the authors described the application of a machine learning technique and method to an orthopedic problem or purpose. By searching both in the Scopus and Medline databases, we retrieved, screened and analyzed the content of 70 journal articles, and coded these resources following an iterative method within a Grounded Theory approach. We report the survey findings by outlining the articles' content in terms of the main machine learning techniques mentioned therein, the orthopedic application domains, the source data and the quality of their predictive performance.

This study aimed to describe the perceptions of decision-makers in major orthopedic centers regarding the value, implementation and use of spine surgery registries. A 33-item survey was sent to CEOs and heads of spine surgery of the International Society of Orthopedic Centers (ISOC). ISOC includes 21 hospitals worldwide with a special focus on high-quality musculoskeletal care. Twelve out of 20 member centers (60%) replied to the survey. Seven have working registries; 5 in Europe and 2 in North America. The estimations for the cost/year were distributed more evenly: $10,000 [2], $20,000 [1], $50,000 [1]. Society cannot afford unnecessary surgery nor renounce to cure patients with effective treatments. Spine surgery registries provide high levels of evidence. The cost of implementing a registry is limited in comparison to RCTs. Spine registries can pragmatically fill our knowledge gap by turning every operated patient into a study participant., Competing Interests: Conflicts of Interest: The authors have no conflicts of interest to declare.

This review focuses on machine learning and on how methods and models combining data analytics and artificial intelligence have been applied to laboratory medicine so far. Although still in its infancy, the potential for applying machine learning to laboratory data for both diagnostic and prognostic purposes deserves more attention by the readership of this journal, as well as by physician-scientists who will want to take advantage of this new computer-based support in pathology and laboratory medicine.

Collecting Patient Reported Outcomes (PROs) is generally seen as an effective way to assess the efficacy and appropriateness of medical interventions, from the patients' perspective. In 2016 the Galeazzi Orthopaedic Institute established a digitized program of PROs collection from spine, hip and knee surgery patients. In this work, we re-port the findings from the data analysis of the responses collected so far about the complementarity of PROs with respect to the data reported by the clinicians, and about the main biases that can undermine their validity and reliability. Although PROs collection is recognized as being far more complex than just asking the patients "how they feel" on a regular basis and it entails costs and devoted electronic platforms, we advocate their further diffusion for the assessment of health technology and clinical procedures.

Evaluation of treatments effectiveness in a context of value-based health care is based on outcomes, and in their assessment. The patient perspective is gaining renovated interest, as demonstrated by the increasing diffusion of Patient Reported Outcome Measure (PROMs) collection initiatives. The concept of Minimal Clinically Important Dif-ference (MID) is generally seen as the basis to estimate the actual effect perceived by the patient after a treatment, like a surgical intervention, but a universally recognized threshold has not yet been established. At the Orthopedic Institute Galeazzi (Milan, Italy) we began a digitized program of PROM collection in spine surgery by means of a digital platform, called Datareg. In this work we aim to investigate MID in the treatment of degenerated disc in terms of patients' perceptions as these are collected through the above electronic registry. We proposed a computation of MID on the basis of two PROM scores, and a critical comparison with a domain expert's proposal.

Purpose: Age-related modifications of tendons, such as reduced tenocyte proliferation and modified extracellular matrix (ECM) turnover, have been previously described, but results are often incomplete and discordant. The aim of this study was to investigate, using morphological and molecular methods, the effect of ageing on human tendons and tenocytes, especially focusing on the collagen turnover pathways, in order to understand how the ageing process could influence tendon biology and structure., Methods: Morphological analysis was performed on fragments from human semitendinosus and gracilis tendons harvested from 10 adult (mean age 41.8 ± 13.3 years) and 6 aged healthy patients (mean age 72.7 ± 7.0 years) by haematoxylin and eosin, Sirius red and Alcian blue staining. The expression of genes and proteins involved in collagen turnover and focal adhesions was assessed by real-time PCR, slot blot and zymography in cultured tenocytes. Cytoskeleton arrangement was studied by immunofluorescence and cell migration by wound healing assay., Results: The structure and composition of ECM in ageing tendons are preserved as well as the expression of genes and proteins involved in collagen turnover pathways. Although morphological analysis revealed that ageing tenocytes tended to an impaired migration potential and that actin filaments are occasionally shorter and randomly distributed, the expression of proteins involved in focal adhesion formation is preserved., Conclusion: Results of this study suggest that the structure of ageing tendons is preserved and that ageing tenocytes maintain their ability for ECM remodelling, supporting the hypothesis that ageing tendons maintain their biomechanical properties. The biological reliability of aged tendons has a clinical relevance, supporting the use of tendon autografts also in the elderly patients. Since the common and successful orthopaedic procedure of anterior cruciate ligament reconstruction using either autografts or allografts is becoming more common in older age groups, these findings suggest that the donor age would not significantly influence the clinical outcome.

Decisional support systems based on machine learning (ML) in medicine are gaining a growing interest as some recent articles have highlighted the high diagnostic accuracy exhibited by these systems in specific medical contexts. However, it is implausible that any potential advantage can be obtained without some potential drawbacks. In light of the current gaps in medical research about the side effects of the application of these new AI systems in medical practice, in this article we summarize the main unexpected consequences that may result from the widespread application of "oracular" systems, that is highly accurate systems that cannot give reasonable explanations of their advice as those endowed with predictive models developed with ML techniques usually are. These consequences range from the intrinsic uncertainty in the data that are used to train and feed these systems, to the inadequate explainability of their output; through the risk of overreliance, deskilling and context desensitization of their end-users. Although some of these issues may be currently hard to evaluate due to the still scarce adoption of these decisional systems in medical practice, we advocate the study of these potential consequences also for a more informed policy of approval beyond hype and disenchantment.

The purpose of this study was to generate consensus among experienced surgeons on "what skills a resident should possess before continuing safe training in the operating room (OR)." An online survey of 65 questions was developed and distributed to surgeons in the European community. A total of 216 responded. The survey included 15 questions regarding generic and specific skills; 16 on patient and tissue manipulation, 11 on knowledge of pathology and 6 on inspection of e-anatomical structures; 5 methods to prepare residents; and 12 on specific skills exercises. The importance of each question (arthroscopic skill) was evaluated ranging from 1 (not important at all) to 6 (very important). Chi-square test, respondent agreement, and a qualitative ranking method were determined to identify the top ranked skills ( p  < 0.05). The top four of general skills considered important were "anatomical knowledge," "tissue manipulation," "spatial perception," and "triangulation" (all chi-square test > 134, p  < 0.001, all excellent agreement > 0.85, and all "high priority" level). The top ranked 2 specific arthroscopic skills were "portal placement" and "triangulating the tip of the probe with a 30-degree scope" (chi-square test > 176, p  < 0.001, excellent agreement, and assigned high priority). The online survey identified consensus on skills that are considered important for a trainee to possess before continuing training in the OR. Compared with the Canadian colleagues, the European arthroscopy community demonstrated similar ranking., Competing Interests: Disclosure The authors report no conflicts of interest in this work., (Thieme Medical Publishers 333 Seventh Avenue, New York, NY 10001, USA.)

The aim of this study was to document the incidence of postoperative shoulder stiffness (SS) after arthroscopic rotator cuff repair and evaluate the role of risk factors for its development. Seventy-five consecutive patients that underwent arthroscopic rotator cuff repair were included. The incidence of postoperative SS was prospectively investigated and the presence of 20 potential risk factors was documented retrospectively. The incidence of post-operative SS was 10.4%. All patients were women, and sex was significantly associated to pathology development (p=0.0067). The presence of gastroesophageal diseases was found to be significantly associated with post-operative SS development (p=0.0046). A significant association between the occurrence of post-operative SS and the presence of gastroesophageal diseases was identified. This finding, not yet reported in literature, deserves further investigation. The incidence of postoperative SS fell among previously reported ranges, with females significantly more affected than men.

In this paper we present two large user studies in which we gather evidence about the adoption and satisfaction level of users in regard to electronic records that manage health related information from two distinct but complementary perspectives: that of General Practitioners (GPs) about their Electronic Medical Records (EMRs); and that of citizens/patients about their Personal Health Records (PHRs). In these user studies we also probe the user attitudes towards innovative functionalities from these two perspectives and, on the basis of the collected perceptions, we apply an original ranking method to infer what features are valued most and hence could inspire design to make PHRs more situated into the users' lives and drive a higher adoption of these tools. On the basis of the perceived shortcomings of current records, we envision an InterPersonal Health Record (IPHR) that is a sort of hybrid electronic record that merges together typical EMR- and PHR-related features and is endowed with specific functionalities aimed at enhancing interpersonal relationships, communication and collaboration between citizens/patients and their GPs through the record and about its contents. This study is then a contribution in understanding the current attitudes and expectations of potential users towards full-fledged prospective PHRs, as well as a first step in identifying those requirements and priority areas on which to focus further for the design and deployment of more community- and communication-oriented tools in the primary health care domain., (Copyright © 2014 Elsevier Ltd. All rights reserved.)

Purpose: To describe preferences and to quantify the amount of agreement among orthopaedic surgeons regarding treatment options for four clinical scenarios of knee pathologies., Methods: A web-based survey was developed to investigate the attitudes of members of an international association of surgeons specialised in sports traumatology and knee surgery European Society of Sports Traumatology, Knee Surgery and Arthroscopy., Results: The response rate was 40 % (412 questionnaires). An inter-rater agreement score (the Normalised Chi-square based Agreement Nomogram, NX2A) was calculated for each question. For scenario 1, 56-year-old male, degenerative medial compartment on both the femoral and tibial side, the surgical approach was preferred to the conservative approach (p < 0.001). Biological procedures were not considered appropriate, and in this respect, the respondents achieved a moderate degree of agreement (NX2A = 0.68). For scenario 2, 35-year-old male, early knee medial arthritis, the surgical treatment was preferred to conservative treatment (p < 0.001). The traditional closed-wedge tibial osteotomy was not regarded as an appropriate treatment with an excellent degree of agreement among surgeons (NX2A = 0.76). For scenario 3, 46-year-old male, ACL lesion, there was an almost disagreement, as respondents showed no preference between a surgical and conservative approach (NX2A = 0.005). Among surgical treatments, an almost perfect agreement with regard to the appropriateness of arthroscopic single-bundle ACL reconstruction with a semitendinosus/gracilis graft was reached by the surgeons (NX2A = 0.8). For scenario 4, 69-year-old male, diffuse knee arthritis (all the compartments), an almost perfect agreement in favour of a total knee prosthesis was obtained for the management of this scenario (NX2A = 0.85)., Conclusions: Web-based survey can help orthopaedic surgeons discuss and propose indications for clinical practice in the management of some of the most common joint diseases., Level of Evidence: Cross-sectional survey, Level V.

Purpose: The aim of this study was to report orthopedic surgeons' management of choice for difficult clinical scenarios of shoulder pathologies., Methods: A web questionnaire was developed including four clinical scenarios of shoulder pathologies. Subsequently, opinions were solicited from more than 1,000 members of an international association of surgeons specialized in sports traumatology and knee surgery (ESSKA)., Results: The response rate was 40% (412 questionnaires). For scenario 1, first anterior dislocation of the shoulder, the most indicated treatment for 71% of respondents was an arthroscopic Bankart repair (P < 0.001). For scenario 2, shoulder arthritis with concentric erosion and cuff tear, 38% chose a shoulder replacement, while 37% preferred a supraspinatus tendon repair in combination with long head of biceps (LHB) tenodesis or tenotomy. For scenario 3, large tendon tears with 70% fatty infiltration of the infraspinatus tendon and lateral LHB instability, 70% of surgeons considered that, among conservative treatments, hyaluronic acid injection was not an appropriate management. Arthroscopic rotator cuff repair, arthroscopic acromioplasty, and LHB tenotomy gained larger consensus (81, 80, and 79% of respondents, respectively). A double-row technique for rotator cuff repair was preferred to a single-row technique (P = 0.02). Scenario 4, adhesive capsulitis, split the respondents equally, with 51% in favor of a surgical approach and 49% in favor of a conservative approach (N.S.)., Conclusions: On-line questionnaires have the potential to improve knowledge about current trends in clinical practice and can help orthopedic surgeons to develop guidelines., Level of Evidence: Cross-sectional; Level V (expert opinion).

In this paper we present the MIR index, a composite index of Medical Information Reliability. This percentage index is intended as a simple indicator of Information Quality of medical online resources from the perspective of their potential consumers. We exemplify its application in the domain of the Non Conventional and Complementary Medicine (NCM/CAM) for its lack of institutional professional figures; yet the structure of the index makes it suitable also for other healthcare-related domains where guidelines for IQ assurance exist.

In this paper we report the findings from a CATI-based survey that involved a representative sample of General Practitioners in Italy. Respondents were asked about their use of ICT in keeping in touch with their patients; their adoption and actual use of their Electronic Patient Record (EPR); and, more importantly, their perception of usefulness of a series of novel functionalities that could enrich the next generation of EPR in general practice.

Hypothesis: This study investigates the rate of infectious and thromboembolic complications in shoulder arthroscopy and their association with pharmacologic prophylaxis., Materials and Methods: On behalf of the Italian Society for Knee Surgery, Arthroscopy, Sport Traumatology, Cartilage and Orthopaedic Technologies (SIGASCOT), we asked the members to complete an on-line Web survey about their experiences and strategies of prophylaxis in shoulder arthroscopy., Results: In the period 2005-2006, 9385 surgeries were performed. We report 15 infections and 6 DVTs. The overall rate of infections was 0.0016 (1.6/1000) and the rate of DVTs was 0.0006 (0.6/1000), Conclusion: The association between infection and antibiotic prophylaxis was significant (P=0.01); however, the risk of DVTs was not decreased with heparin prophylaxis., Level of Evidence: Level 3.

Artificial rainfalls were used to determine the effect of the amount of the rainfall and the time interval between pesticide application and rainfall event, on folpet and mancozeb residues on grapes and vine leaves. Forty-five mm of rain were administered to the vineyard in different amounts (45; 30+15; 15+15+15 mm). Folpet showed good rainfastness on the grapes and on the leaves. A modest decrease was observed only in the experiments that had received 45 mm of rain at one go. Mancozeb showed a lower rainfastness, since a portion of the deposit was easily washed off also by a modest rainfall. The percentage of this portion was higher in the grapes (38%) than in the leaves (20%). The data obtained in these experiments show that, in the case of folpet, it is not necessary to repeat the treatment when it rains the day after, while it is recommendable to repeat it in the case of mancozeb.

The behaviour of the fungicide, fenhexamid, on grapes, and during wine-making, as well as its effect on the microflora of alcoholic and malolactic fermentation has been studied. After treatment, the residue on grapes decreased rapidly to one-third of the initial level after the first week, while it remained constant during the following two weeks. At harvest, in the wine obtained by vinification without skins, the fungicide residue decreased on average by 49%, while in the wine obtained by vinification with skins, the decrease was on average 62%. The presence of this fungicide on grapes and in the wine did not affect alcoholic and malolactic fermentation, nor did fermentation cause any degradation of the fungicide. A simple and rapid gas chromatograhic method (GC-NPD) for the determination of fenhexamid residues in grapes, must and wine is described.

Quinoxyfen is a new fungicide that belongs to the family of the quinolines, recently introduced to control powdery mildew (Uncinula necator). In this paper the fate of quinoxyfen residues from vine to wine and in their processing products was studied. After the last of four applications at the recommended rate, 0.38 mg/kg of residue was found on the grapes, which is under the legal limit fixed in Italy (0.5 mg/kg). The degradation rate was according to a pseudo-first-order kinetics (r = 0.964) and the half-life was 7.24 days. Vinification was carried out with and without maceration. During the vinifications without maceration <50% of the residues passed from the grapes to the musts. Separation of the lees (8%) from the must by centrifugation caused no detectable residues in centrifuged must. At the end of fermentation with and without maceration no quinoxyfen residues were determinable in the wine. No effect on the alcoholic or malolactic fermentation was observed even in the presence of higher quinoxyfen concentrations than those found in the grapes at harvest time. During fermentation, the yeasts partially degraded the pesticides and completly adsorbed them. Bacteria, on the other hand, do not have any degradative effect on the pesticides. The raisins obtained by sun-drying did not contain any residues, whereas those obtained by oven-drying show the same amount of residues as in the fresh grapes. During the sun-drying process the fruit weight decreased by a factor of 4; the decrease in the oven-drying was equivalent. Samples of dregs and liquid lees, fortified with high levels of quinoxyfen. were double-distilled. The first dregs distillate, with an alcohol content of 32.1%, did not show any residues, whereas the first lees distillate, with an alcohol content of 34.5%, showed 7% of the initial residues. After the second lees distillation, the obtained product showed an alcoholic content of 81.2% and no residues of quinoxyfen (<0.01 mg/kg).

Field trials were carried out to study the persistence of acephate and buprofezin on olives. Two cultivars, pizz'e carroga and pendolino, with very large and small fruits respectively were used. After treatment, no difference was found between the two pesticide deposits on the olives. The disappearance rates, calculated as pseudo first order kinetics, were similar for both pesticides (on average 12 days). Methamidophos, the acephate metabolite, was always present on all olives, and in some pendolino samples it showed higher residues than the maximum residue limit (MRL). During washing, the first step of olive processing, the residue level of both pesticides on the olives did not decrease. After processing of the olives into oil, no residues of acephate or methamidophos were found in the olive oil, while the residues of buprofezin were on average four times higher than on olives.

Field trials were carried out to evaluate whether folpet sprayed on grapevines penetrated the epicuticular wax and cell walls of grapes. Folpet showed poor penetration into the epicuticular wax; it was found almost totally on the surface. Despite its low solubility in water, perhaps due to the presence of adjuvants, its residues showed such a high resistance to washing that the action of rain was negligible in decreasing residues.

A cross-sectional study of tooth maturation was carried out in a group of 382 healthy children (178 boys and 204 girls) from Sardinia. A single orthopantomogram was examined for each patient and the development of permanent mandibular canines, premolars, and molars was evaluated according to Moorrees' method [Moorrees CFA, Fanning EA, Hunt EE: J Dent Res 42:1490-1502, 1963] Dental maturity scores and chronological ages were processed by linear regression analysis. Significant differences were observed in the timing of tooth development between boys and girls. A visual comparison with North American children revealed appreciable differences between the two populations, accounting for a delayed dental maturation in Sardinian children.