Polymethyl Methacrylate for Elbow Arthroplasty: Is There Another Way?

Polymethyl methacrylate remains the only US Food and Drug Administration-approved method of total elbow arthroplasty fixation and exhibits high aseptic loosening rates that result in challenging revision surgeries and potential morbidity secondary to bone-cement implantation syndrome. In this policy paper, the authors aim to explore the historical background of polymethyl methacrylate and the complications that arise in association with its use. We will review arthroplasty trends in the elbow and lower extremities and the challenges with the US Food and Drug Administration-approval process.

Establishing regions of interest of the lower leg and ankle for perioperative volumetric assessment with a portable 3D scanner in orthopedic and trauma surgery - a pilot study.

BACKGROUND: Soft tissue swelling assessment benefits from a reproducible and easy to use measurement method. Monitoring of the injured lower extremity is of clinical import during staged soft tissue management. Portable 3D scanners offer a novel and precise option to quantify and contrast the shapes and volumes of the injured and contralateral uninjured limbs. This study determined three regions of interest (ROI) within the lower extremity (lower leg, ankle and foot), that can be used to evaluate 3D volumetric assessment for staged soft tissue management in orthopedic and trauma surgery. METHODS: Twelve healthy volunteers (24 legs) were included in this cohort study. Scans of all three ROI were recorded with a portable 3D scanner (Artec, 3D scanner EVA) and compared between the right and left leg using the software Artec Studio (Arctec Group, Luxemburg). RESULTS: Mean volume of the right leg was 1926.64 ± 308.84 ml (mean ± SD). ROI: lower leg: 931.86 ± 236.15 ml; ankle: 201.56 ± 27.88 ml; foot: 793.21 ± 112.28 ml. Mean volume of the left leg was 1937.73 ± 329.51 ml. ROI: lower leg: 933.59 ± 251.12 ml; ankle: 201.53 ± 25.54 ml; foot: 802.62 ± 124.83 ml. There was no significant difference of the overall volume between right and left leg (p > 0.05; overall volume: △ difference: 29.5 ± 7.29 ml, p = 0.8; lower leg: △ difference: 21.5 ± 6.39 ml, p = 0.8; ankle: △ difference: 5.3 ± 2.11 ml, p = 0.4; △ difference: 16.33 ± 4.45 ml, p = 0.8. CONCLUSION: This pilot study defines three regions of interest of the lower leg and demonstrates no difference between the right and left side. Based on these ROI, further studies are needed to evaluate the clinical applicability of the scanner.

Volumetry of Hand and Forearm: A 3D Volumetric Approach.

BACKGROUND: Swelling and edema of the hand and forearm may occur in various traumatic and degenerative diseases. So far, no precise measurement protocol exists. The objective of this study was to evaluate an examination protocol with relevant regions of interest (ROIs) measured by a 3-dimensional (3D) scanner to achieve precise, reproducible, and objective measurements for an optimized detection of volumes of the hand and forearm. METHODS: A 3D scan protocol was developed using an Artec, 3D scanner EVA to measure discrete hand volumes of healthy volunteers. Five areas were defined as ROIs, representing volumes of the finger, metacarpus, wrist, hand, and distal forearm. Contralateral limbs were used for volume comparisons and calculation of volume differences. RESULTS: For this study, 12 individuals (58.3% women, 24 hands and forearms) with a mean age of 27.1 ± 3 years were included. Mean volume values for left and right ROIs correlated with each other, with slightly higher volumes for the right upper extremity. Volume differences showed statistically significant results for the finger region (ROI I; P = .009), the metacarpal region (ROI II; P < .001), hand region (ROI IV; P = .001), and forearm region (ROI V; P = .006), with the exception of the wrist region (ROI III; P = .722). CONCLUSIONS: Our results demonstrate that this 3D volumetric approach is a reliable and objective tool for measuring volumes and circumferences in hand and forearm. Based on our determined ROIs, further studies are needed to explore the significance for clinical applications.

Barriers and Facilitators to the Implementation of Family-Centered Technology in Complex Care: Feasibility Study.

BACKGROUND: Care coordination is challenging but crucial for children with medical complexity (CMC). Technology-based solutions are increasingly prevalent but little is known about how to successfully deploy them in the care of CMC. OBJECTIVE: The aim of this study was to assess the feasibility and acceptability of GoalKeeper (GK), an internet-based system for eliciting and monitoring family-centered goals for CMC, and to identify barriers and facilitators to implementation. METHODS: We used the Consolidated Framework for Implementation Research (CFIR) to explore the barriers and facilitators to the implementation of GK as part of a clinical trial of GK in ambulatory clinics at a children's hospital (NCT03620071). The study was conducted in 3 phases: preimplementation, implementation (trial), and postimplementation. For the trial, we recruited providers at participating clinics and English-speaking parents of CMC<12 years of age with home internet access. All participants used GK during an initial clinic visit and for 3 months after. We conducted preimplementation focus groups and postimplementation semistructured exit interviews using the CFIR interview guide. Participant exit surveys assessed GK feasibility and acceptability on a 5-point Likert scale. For each interview, 3 independent coders used content analysis and serial coding reviews based on the CFIR qualitative analytic plan and assigned quantitative ratings to each CFIR construct (-2 strong barrier to +2 strong facilitator). RESULTS: Preimplementation focus groups included 2 parents (1 male participant and 1 female participant) and 3 providers (1 in complex care, 1 in clinical informatics, and 1 in neurology). From focus groups, we developed 3 implementation strategies: education (parents: 5-minute demo; providers: 30-minute tutorial and 5-minute video on use in a clinic visit; both: instructional manual), tech support (in-person, virtual), and automated email reminders for parents. For implementation (April 1, 2019, to December 21, 2020), we enrolled 11 providers (7 female participants, 5 in complex care) and 35 parents (mean age 38.3, SD 7.8 years; n=28, 80% female; n=17, 49% Caucasian; n=16, 46% Hispanic; and n=30, 86% at least some college). One parent-provider pair did not use GK in the clinic visit, and few used GK after the visit. In 18 parent and 9 provider exit interviews, the key facilitators were shared goal setting, GK's internet accessibility and email reminders (parents), and GK's ability to set long-term goals and use at the end of visits (providers). A key barrier was GK's lack of integration into the electronic health record or patient portal. Most parents (13/19) and providers (6/9) would recommend GK to their peers. CONCLUSIONS: Family-centered technologies like GK are feasible and acceptable for the care of CMC, but sustained use depends on integration into electronic health records. TRIAL REGISTRATION: ClinicalTrials.gov NCT03620071; https://clinicaltrials.gov/ct2/show/NCT03620071.

Defining the region of interest of the knee for perioperative volumetric assessment with a portable 3D scanner in orthopedic and trauma surgery.

BACKGROUND: The aim of this study was to characterize three regions of interest (ROI) around the knee with a portable 3D scanner (Artec 3D scanner EVA). Soft tissue topography assessment with an optimized, precise, and reproducible method may assist surgeons when managing soft tissue swelling in the post traumatic setting. METHODS: 12 healthy volunteers (24 legs, 7 women, 5 man) were included in this study. The patient cohort showed a mean age of 27.1 years (SD±3), a mean weight of 70 kg (SD±13) and a mean height of 171 cm (SD±8.8). All scans were recorded by the same examiner in the same room and with the same scanner (Artec, 3 D scanner EVA). Three volume regions of interest (ROI) were defined: the distal femur (circumference measured between the of superior extent of the patella to 10 cm proximal), the knee joint (measured from the top of the patella to the tibial tuberosity) and the proximal tibia (tibial tuberosity to 10 cm distal). RESULTS: The mean volume of the right leg was 3.901 l (I. distal femur: 1.63 l, knee joint: 1.33 l, proximal tibia: 1.10 l) and mean volume of the left leg was 3.910 l (I. distal femur: 1.66 l, knee joint: 1.34 l, proximal tibia: 1.12 l). The volume difference between the right and left leg was 0.094 l (SD ± 0.083 l) The Wilcoxon-Mann-Whitney test showed no significant differences of the volumes between the right and left leg. CONCLUSIONS: This study demonstrates that portable 3D scanning could be an accurate and reliable tool for orthopedics and trauma surgeons. Based on the ROIs of this pilot study, further studies are needed to test the significance for clinical applications for patients with an injured knee.

Metacarpal Arc Motion: Comparison of Different Measurement Methods.

Introduction The range of motion of the carpometacarpal (CMC) joints is difficult to assess in clinical practice. Materials and Methods The purpose of this study was to measure a precise range of motion measurement of the metacarpal arc during maximum palmar flexion of the fifth CMC joint in healthy persons with three different methods: optical, electrogoniometrical, and sonographical. Results The metacarpal arc motion measured with the optical technique was 24.9 degrees (standard deviation [SD]: 6.2 degrees), with the electrogoniometric technique was 24.3 degrees (SD: 3.6 degrees) and with sonography was 25.1 degrees (SD: 3.5 degrees), with no differences between left and right hand for all methods. Conclusion This study shows that different techniques can be used for the measurement of the metacarpal arc motion. With these techniques, it is possible to assess the metacarpal arc in patients with fractures to the CMC joints to evaluate posttraumatic function und rehabilitation progress. This study shows level of evidence as III.

Accelerated antimicrobial discovery via deep generative models and molecular dynamics simulations.

The de novo design of antimicrobial therapeutics involves the exploration of a vast chemical repertoire to find compounds with broad-spectrum potency and low toxicity. Here, we report an efficient computational method for the generation of antimicrobials with desired attributes. The method leverages guidance from classifiers trained on an informative latent space of molecules modelled using a deep generative autoencoder, and screens the generated molecules using deep-learning classifiers as well as physicochemical features derived from high-throughput molecular dynamics simulations. Within 48 days, we identified, synthesized and experimentally tested 20 candidate antimicrobial peptides, of which two displayed high potency against diverse Gram-positive and Gram-negative pathogens (including multidrug-resistant Klebsiella pneumoniae) and a low propensity to induce drug resistance in Escherichia coli. Both peptides have low toxicity, as validated in vitro and in mice. We also show using live-cell confocal imaging that the bactericidal mode of action of the peptides involves the formation of membrane pores. The combination of deep learning and molecular dynamics may accelerate the discovery of potent and selective broad-spectrum antimicrobials.

Parent Perspectives in Shared Decision-Making for Children With Medical Complexity.

OBJECTIVE: Shared decision-making (SDM) may improve outcomes for children with medical complexity (CMC). CMC have lower rates of SDM than other children, but little is known about how to improve SDM for CMC. The objective of this study is to describe parent perspectives of SDM for CMC and identify opportunities to improve elements of SDM specific to this vulnerable population. METHODS: Interviews with parents of CMC explored SDM preferences and experiences. Eligible parents were ≥18 years old, English- or Spanish-speaking, with a CMC <12 years old. Interviews were recorded, transcribed, and analyzed by independent coders for shared themes using modified grounded theory. Codes were developed using an iterative process, beginning with open-coding of a subset of transcripts followed by discussion with all team members, and distillation into preliminary codes. Subsequent coding reviews were conducted until no new themes emerged and existing themes were fully explored. RESULTS: We conducted interviews with 32 parents (27 in English, mean parent age 34 years, standard deviation = 7; mean child age 4 years, standard deviation = 4; 50% with household income <$50,000, 47% with low health literacy) in inpatient and outpatient settings. Three categories of themes emerged: participant, knowledge, and context. Key opportunities to improve SDM included: providing a shared decision timeline, purposefully integrating patient preferences and values, and addressing uncertainty in decisions. CONCLUSION: Our results provide insight into parent experiences with SDM for CMC. We identified unique opportunities to improve SDM for CMC that will inform future research and interventions to improve SDM for CMC.

Assessment of Articular Cartilage Disorders After Distal Radius Fracture Using Biochemical and Morphological Nonenhanced Magnetic Resonance Imaging.

PURPOSE: To assess radiocarpal articular cartilage after distal radius fracture, with and without intra-articular extension, compared with healthy controls using multiparametric, nonenhanced magnetic resonance imaging (MRI). METHODS: In this prospective study, multiparametric MRI of the radiocarpal articular cartilage was performed in 26 participants (16 males and 10 females; mean age, 39.5 ± 14.7 years; range, 20-70 years) using 3T MRI. The cohort consisted of 14 patients with a distal radius fracture and 12 healthy volunteers. The radiocarpal articular cartilage was assessed using morphological (Double Echo Steady-State [DESS] and True Fast Imaging With Steady-State Precession [TrueFISP]) and biochemical (T2∗) MRI sequences without an intravenous contrast agent. The modified Outerbridge classification system for morphological analyses and region-of-interest biochemical analysis were applied to assess the degree of articular cartilage damage in each patient. RESULTS: Morphological articular cartilage assessment showed no difference between the DESS sequence and the reference standard, TrueFISP. In the morphological (DESS and TrueFISP) and biochemical (T2∗) assessments, patients with intra-articular fractures did not show articular cartilage damage different from those with extra-articular fractures. Greater articular cartilage degradation was observed after distal radius fracture compared with controls. CONCLUSIONS: Posttraumatic radiocarpal articular cartilage damage did not differ between fractures with intra-articular and extra-articular extension, but patients with fractures had notably higher articular cartilage degradation compared with healthy controls. Magnetic resonance imaging using advanced multiparametric sequences may facilitate accurate, noninvasive assessment of articular cartilage changes after distal radius fracture without the need for a contrast agent. TYPE OF STUDY/LEVEL OF EVIDENCE: Diagnostic IV.

Performance under pressure in skill tasks: An analysis of professional darts.

Understanding and predicting how individuals perform in high-pressure situations is of importance in designing and managing workplaces. We investigate performance under pressure in professional darts as a near-ideal setting with no direct interaction between players and a high number of observations per subject. Analyzing almost one year of tournament data covering 32,274 dart throws, we find no evidence in favor of either choking or excelling under pressure.

Doctor, when can I drive? - Range of functional ankle motion during driving.

BACKGROUND: Driving a motor vehicle needs a specific joint mobility and yet only limited knowledge exists regarding the necessary ankle range of motion. The goal of this study is to characterize the sequence and range of ankle motion. METHODS: The arc of plantarflexion/dorsiflexion and supination/pronation was recorded in the right and left ankle using electrogoniometers while thirty laps were driven by fifteen healthy participants around a course in a manual transmission car with a left sided steering wheel. The driver was required to perform the following maneuvers during each lap: (I) Vehicle acceleration and gear change, (II) Sudden evasion, (III) Routine turning, (IV) Rapid turning, (V) Vehicle acceleration followed by emergency braking. RESULTS: Driving required the right ankle to plantarflex 13±9 and dorsiflex 22±7 while supinating 15±7 degrees and pronating minimally. The left ankle plantarflexed 19±10and dorsiflexed 17±10 while supinating 15±7 degrees and pronating minimally. The right ankle dorsiflexed significantly more (p=0.00), and yet the left ankle had a significantly higher maximum plantarflexion and range of plantarflexion/dorsiflexion (p=0.00). Emergency braking resulted in a significantly higher maximum plantarflexion as well as plantarflexion/dorsiflexion range when compared to other maneuvers. CONCLUSION: This study describes the range of ankle motion identified to drive a car with a manual transmission and a left-sided steering wheel. The right and left ankle exhibit different arcs of motion during driving. This knowledge may assist when evaluating a patient's driving capability. Further studies are needed to investigate whether movement restrictions impair driving. LEVEL OF EVIDENCE: Basic science study.

Visual Interaction with Deep Learning Models through Collaborative Semantic Inference.

Automation of tasks can have critical consequences when humans lose agency over decision processes. Deep learning models are particularly susceptible since current black-box approaches lack explainable reasoning. We argue that both the visual interface and model structure of deep learning systems need to take into account interaction design. We propose a framework of collaborative semantic inference (CSI) for the co-design of interactions and models to enable visual collaboration between humans and algorithms. The approach exposes the intermediate reasoning process of models which allows semantic interactions with the visual metaphors of a problem, which means that a user can both understand and control parts of the model reasoning process. We demonstrate the feasibility of CSI with a co-designed case study of a document summarization system.

Doctor, when can I drive?-the range of elbow motion while driving a car.

BACKGROUND: Immobilization of the upper extremity after an acute injury or postoperatively affects an individual's ability to safely operate a motor vehicle. The elbow is particularly sensitive to immobilization, with subsequent stiffness leading to functional limitations. Most activities of daily living are successfully achieved within a "functional arc" of elbow motion between 30° and 130° of flexion. No objective guidelines exist regarding the range of motion needed to safely operate a vehicle. In this study, we measured the range of motion of right and left elbows while driving a manual-transmission car. MATERIALS AND METHODS: Using electro-goniometers, we measured the flexion and extension, as well as pronation and supination, of the right and left elbows in 20 healthy, right hand-dominant subjects while driving a car. These measurements were recorded on (1) city streets, (2) country roads, and (3) highways. RESULTS: For city streets, the range of motion in terms of flexion and pronation/supination was 15°-105° and 0°-45°/0°-35°, respectively, for the right elbow and 20°-95° and 0°-45°/0°-40°, respectively, for the left. For country roads, it was 10°-100° and 0°-40°/0°-35°, respectively, for the right elbow and 20°-95° and 0°-30°/0°-30°, respectively, for the left. For highways, it was 5°-100° and 0°-40°/0°-35°, respectively, for the right elbow and 20°-90° and 0°-30°/0°-25°, respectively, for the left. Mean pronation was significantly higher for the right elbow (P < .01). CONCLUSION: This study describes the range of elbow motion identified to drive a car with a manual transmission and a left-sided steering wheel. Mean pronation of the right elbow is significantly higher than that of the left. Further studies are needed to investigate the relevance of movement restrictions as they relate to handedness, steering-wheel side, and driving impairment.

Influence of Wrist Position on the Metacarpophalangeal Joint Motion of the Index Through Small Finger.

BACKGROUND: The metacarpophalangeal joints exhibit range of motion that is influenced by wrist position. Synergistic motion occurs between the wrist and the metacarpophalangeal joints with different static wrist positions affecting joints' motion capability. The aim of this study was to determine how different wrist positions influence the active range of motion of the index through small finger metacarpophalangeal joints. METHODS: The active range of motion of the index through small finger metacarpophalangeal joints of 31 healthy subjects was measured in flexion/extension and radial/ulnar deviation in 5 different flexion/extension wrist positions, using biaxial electrogoniometers. RESULTS: There was a difference in range of motion of all fingers depending on the wrist position. The minimum metacarpophalangeal joint range of motion was found in 80° wrist extension, the maximum in neutral wrist position. For the index finger, flexion/extension was 84.7° (±8.6°) to 25.9° (±10.2°) and radial/ulnar deviation was 32.1° (±11.3°) to 22.6° (±12.8°). For the middle finger, flexion/extension was 84.8° (±8.5°) to 25.9° (±10.1°) and radial/ulnar deviation 28.8° (±11.1°) to 22.1° (±8.9). The fourth finger showed a range of motion for flexion/extension of 87.2° (±11.5°) to 22.8° (±11.6°) and radial/ulnar deviation of 8.1° (±5.8°) to 32.3° (±12.4°). The highest range of motion was measured at the fifth finger with flexion/extension of 84.0° (±8.6°) to 32.1°(±16.8°) and radial/ulnar deviation of 15.1° (±12.9°) up to 54.6° (±18.7°). CONCLUSIONS: The range of motion of the index through small finger metacarpophalangeal joints was significantly influenced by wrist position. The highest metacarpophalangeal joint range of motion of all fingers was conducted in neutral wrist positions. Apart from ergonomic implications, we conclude that metacarpophalangeal joint motion should be assessed under standardized wrist positions.

Quantifying the Center of Elbow Rotation: Implications for Medial Collateral Ligament Reconstruction.

BACKGROUND: Medial collateral ligament (MCL) reconstruction of the elbow mandates precise characterization of where the centerline of elbow rotation projects onto the medial epicondyle (ME). A muscle-splitting approach allows the flexor-pronator muscles to remain attached to the ME and facilitates visualization of the MCL remnant, the sublime tubercle, and the ulnohumeral joint line. Knowledge of where the centerline of rotation intersects the ME relative to the ulnohumeral joint line may assist the surgeon during placement of the proximal drill hole. METHODS: Models were created from the computed tomography scans of 29 normal elbows. The centerline of rotation, center of the trochlea, sublime tubercle, and ulnohumeral joint line were identified. Measurements were taken from the ulnohumeral joint line to the center of the trochlea and to the centerline of rotation in the sagittal view and along the course of the MCL. RESULTS: The centerline of rotation intersected the ME in a consistent location. With the elbow flexed 90°, the trochlea center and the centerline of rotation are essentially in line with each other. There are significant differences between the distances from the ulnohumeral joint line to the center of the trochlea and to the centerline of rotation in both the sagittal view and along the course of the MCL. CONCLUSIONS: The centerline of rotation is located 14.31 mm (1.70) from the ulnohumeral joint line in the sagittal view and 16.54 mm (2.09) from the ulnohumeral joint line along the course of the MCL.

Differences of radiocarpal cartilage alterations in arthritis and osteoarthritis using morphological and biochemical magnetic resonance imaging without gadolinium-based contrast agent administration.

OBJECTIVES: To identify differences of radiocarpal cartilage alterations in osteoarthritis and arthritis using multiparametrical magnetic resonance imaging (MRI) comprising morphological and biochemical sequences without gadolinium-based contrast agent administration. METHODS: In this prospective study, multiparametrical MRI of the radiocarpal cartilage was performed in 47 participants (mean age, 46.6 ± 17.6 years; min., 20 years; max., 79 years) on a 3 Tesla MRI. The cohort consisted of 11 patients suffering from arthritis, 10 patients with osteoarthritis, 14 patients after distal radius fracture, and 12 healthy volunteers. The radiocarpal cartilage was assessed using morphological (DESS, TrueFISP) and biochemical (T2*) MRI sequences without the application of intravenous contrast agent. The modified Outerbridge classification system for morphological and region-of-interest analyses for biochemical analysis was applied to assess the degree of cartilage damage in each patient before data were statistically tested for significant difference between the groups using a post hoc Tukey test. RESULTS: In morphological imaging, cartilage damage was significantly more frequent in arthritis and osteoarthritis than in healthy volunteers (DESS: p = 0.01, p = 0.0004; TrueFISP: p = 0.02, p = 0.0001). In T2* imaging, patients with osteoarthritis showed higher cartilage damage compared to patients with arthritis (p = 0.01). CONCLUSION: With multiparametrical MRI, it is possible to identify differences of radiocarpal cartilage alterations of patients with arthritis and osteoarthritis using the combination of morphological and biochemical MR imaging of the radiocarpal cartilage without the application of contrast agent. Multiparametrical MRI without the usage of contrast agent may be a potential tool helping to distinguish both entities. KEY POINTS: • Multiparametrical MRI with morphological and biochemical sequences allows the differentiation of patients with arthritis and osteoarthritis. • High-resolution MRI of radiocarpal cartilage is possible without administration of contrast agent.

Doctor, when can I drive? - Range of motion of the knee while driving a car.

BACKGROUND: One of the most important activities of daily living is operating a motor vehicle. With increasing age the prevalence of musculoskeletal disorders such as knee osteoarthritis may interfere with an individual's ability to do so safely. Physicians are tasked with determining a patient's ability to drive and yet the necessary joint range of motion required for driving a car has not been characterized. METHODS: The range of motion of the right and left knees was recorded using electrogoniometers in 20 healthy subjects while driving a car on three route types (a) city streets, b) country roads and c) highways). Special emphasis was placed on the left knee associated with changing a gear. RESULTS: The range of motion while driving is 40-80° for the right and 20-85° flexion for the left knee. A significant difference was noted for each side (p < 0.01) with a higher flexion occurring in the left knee (p < 0.01). The average position of the knee while changing a gear (beginning, maximum, ending) was: right: 55°±10°, 62°±10°, 53°±10°; left: 67°±7°, 39°±8°, 66°±8° (mean flexion±standard deviation). CONCLUSION: This study characterized the knee range of motion that occurs while driving a car. Our data suggests that common driving activities such as accelerating a vehicle or braking can be achieved with the right knee through a limited range of motion. The greater range of motion and the higher flexion of the left knee are mainly attributed to the gear changing. The present data may benefit physicians in their evaluation of driving capability.

SEQ2SEQ-VIS : A Visual Debugging Tool for Sequence-to-Sequence Models.

Neural sequence-to-sequence models have proven to be accurate and robust for many sequence prediction tasks, and have become the standard approach for automatic translation of text. The models work with a five-stage blackbox pipeline that begins with encoding a source sequence to a vector space and then decoding out to a new target sequence. This process is now standard, but like many deep learning methods remains quite difficult to understand or debug. In this work, we present a visual analysis tool that allows interaction and "what if"-style exploration of trained sequence-to-sequence models through each stage of the translation process. The aim is to identify which patterns have been learned, to detect model errors, and to probe the model with counterfactual scenario. We demonstrate the utility of our tool through several real-world sequence-to-sequence use cases on large-scale models.

Behind the scenes: A medical natural language processing project.

Advancement of Artificial Intelligence (AI) capabilities in medicine can help address many pressing problems in healthcare. However, AI research endeavors in healthcare may not be clinically relevant, may have unrealistic expectations, or may not be explicit enough about their limitations. A diverse and well-functioning multidisciplinary team (MDT) can help identify appropriate and achievable AI research agendas in healthcare, and advance medical AI technologies by developing AI algorithms as well as addressing the shortage of appropriately labeled datasets for machine learning. In this paper, our team of engineers, clinicians and machine learning experts share their experience and lessons learned from their two-year-long collaboration on a natural language processing (NLP) research project. We highlight specific challenges encountered in cross-disciplinary teamwork, dataset creation for NLP research, and expectation setting for current medical AI technologies.