Approaching Artificial Intelligence in Orthopaedics: Predictive Analytics and Machine Learning to Prognosticate Arthroscopic Rotator Cuff Surgical Outcomes.

Machine learning (ML) has not yet been used to identify factors predictive for post-operative functional outcomes following arthroscopic rotator cuff repair (ARCR). We propose a novel algorithm to predict ARCR outcomes using machine learning. This is a retrospective cohort study from a prospectively collected database. Data were collected from the Surgical Outcome System Global Registry (Arthrex, Naples, FL, USA). Pre-operative and 3-month, 6-month, and 12-month post-operative American Shoulder and Elbow Surgeons (ASES) scores were collected and used to develop a ML model. Pre-operative factors including demography, comorbidities, cuff tear, tissue quality, and fixation implants were fed to the ML model. The algorithm then produced an expected post-operative ASES score for each patient. The ML-produced scores were compared to actual scores using standard test-train machine learning principles. Overall, 631 patients who underwent shoulder arthroscopy from January 2011 to March 2020 met inclusion criteria for final analysis. A substantial number of the test dataset predictions using the XGBoost algorithm were within the minimal clinically important difference (MCID) and substantial clinical benefit (SCB) thresholds: 67% of the 12-month post-operative predictions were within MCID, while 84% were within SCB. Pre-operative ASES score, pre-operative pain score, body mass index (BMI), age, and tendon quality were the most important features in predicting patient recovery as identified using Shapley additive explanations (SHAP). In conclusion, the proposed novel machine learning algorithm can use pre-operative factors to predict post-operative ASES scores accurately. This can further supplement pre-operative counselling, planning, and resource allocation. Level of Evidence: III.

Streamlining the KOOS Activities of Daily Living Subscale Using Machine Learning.

Background: Functional outcome scores provide valuable data, yet they can be burdensome to patients and require significant resources to administer. The Knee injury and Osteoarthritis Outcome Score (KOOS) is a knee-specific patient-reported outcome measure (PROM) and is validated for anterior cruciate ligament (ACL) reconstruction outcomes. The KOOS requires 42 questions in 5 subscales. We utilized a machine learning (ML) algorithm to determine whether the number of questions and the resultant burden to complete the survey can be lowered in a subset (activities of daily living; ADL) of KOOS, yet still provide identical data. Hypothesis: Fewer questions than the 17 currently provided are actually needed to predict KOOS ADL subscale scores with high accuracy. Study Design: Cohort study (diagnosis); Level of evidence, 2. Methods: Pre- and postoperative patient-reported KOOS ADL scores were obtained from the Surgical Outcome System (SOS) data registry for patients who had ACL reconstruction. Categorical Boosting (CatBoost) ML models were built to analyze each question and its value in predicting the patient's actual functional outcome (ie, KOOS ADL score). A streamlined set of minimal essential questions were then identified. Results: The SOS registry contained 6185 patients who underwent ACL reconstruction. A total of 2525 patients between the age of 16 and 50 years had completed KOOS ADL scores presurgically and 3 months postoperatively. The data set consisted of 51.84% male patients and 48.16% female patients, with a mean age of 29 years. The CatBoost model predicted KOOS ADL scores with high accuracy when only 6 questions were asked (R2 = 0.95), similar to when all 17 questions of the subscale were asked (R2 = 0.99). Conclusion: ML algorithms successfully identified the essential questions in the KOOS ADL questionnaire. Only 35% (6/17) of KOOS ADL questions (descending stairs, ascending stairs, standing, walking on flat surface, putting on socks/stockings, and getting on/off toilet) are needed to predict KOOS ADL scores with high accuracy after ACL reconstruction. ML can be utilized successfully to streamline the burden of patient data collection. This, in turn, can potentially lead to improved patient reporting, increased compliance, and increased utilization of PROMs while still providing quality data.

The effect of antiapoptosis genes on clarification performance.

Optimal bioreactor harvest time is typically determined based on maximizing product titer without compromising product quality. We suggest that ease of downstream purification should also be considered during harvest. In this view, we studied the effect of antiapoptosis genes on downstream performance. Our hypothesis was that more robust cells would exhibit less cell lysis and thus generate lower levels of cell debris and host-cell contaminants. We focused on the clarification unit operation, measuring postclarification turbidity and host-cell protein (HCP) concentration as a function of bioreactor harvest time/cell viability. In order to mimic primary clarification using disk-stack centrifugation, a scale-down model consisting of a rotating disk (to simulate shear in the inlet feed zone of the centrifuge) and a swinging-bucket lab centrifuge was used. Our data suggest that in the absence of shear during primary clarification (typical of depth filters), a 20-50% reduction in HCP levels and 50-65% lower postcentrifugation turbidity was observed for cells with antiapoptosis genes compared to control cells. However, on exposing the cells to shear levels typical in a disk-stack centrifuge, the reduction in HCP was 10-15% while no difference in postcentrifugation turbidity was observed. The maximum benefit of antiapoptosis genes is, therefore, realized using clarification options that involve low shear, <1 × 10(6) W/m(3) and minimal damage to the cells.

Biophysical characterization of DNA and RNA aptamer interactions with hen egg lysozyme.

This work characterized the binding of an RNA aptamer recognizing hen egg white lysozyme, as well as a literature-reported single-stranded DNA analog of sequence identical to the original RNA aptamer, using fluorescence anisotropy, isothermal titration calorimetry (ITC) and analytical ultracentrifugation. The polyanionic DNA aptamer analog is selective for lysozyme even over cationic cytochrome c and has been reported to be successfully used in biosensing applications. The association however, is predominantly of electrostatic character, strongly salt-sensitive and entropically-driven, in contrast to previously described enthalpically-driven antibody-lysozyme and DNA aptamer-VEGF interactions. With a moderate selectivity for their target, high salt-sensitivity along with fast association and dissociation behavior, these molecules might serve as pseudo-affinity ligands for biomolecular separations.

Engineered 5S ribosomal RNAs displaying aptamers recognizing vascular endothelial growth factor and malachite green.

In previous work, Vibrio proteolyticus 5S rRNA was shown to stabilize 13-50 nucleotide "guest" RNA sequences for expression in Escherichia coli. The expressed chimeric RNAs accumulated to high levels in E. coli without being incorporated into ribosomes and without obvious effects on the host cells. In this work, we inserted sequences encoding known aptamers recognizing a protein and an organic dye into the 5S rRNA carrier and showed that aptamer function is preserved in the chimeras. A surface plasmon resonance competitive binding assay demonstrated that a vascular endothelial growth factor (VEGF) aptamer/5S rRNA chimera produced in vitro by transcriptional runoff could compete with a DNA aptamer for VEGF, implying binding of the growth factor by the VEGF "ribosomal RNA aptamer." Separately, a 5S rRNA chimera displaying an aptamer known to increase the fluorescence of malachite green (MG) also enhanced MG fluorescence. Closely related control rRNA molecules showed neither activity. The MG aptamer/5S rRNA chimera, like the original MG aptamer, also increased the fluorescence of other triphenyl methane (TPM) dyes such as crystal violet, methyl violet, and brilliant green, although less effectively than with MG. These results indicate that the molecular recognition properties of aptamers are not lost when they are expressed in the context of a stable 5S rRNA carrier. Inclusion of the aptamer in a carrier may facilitate production of large quantities of RNA aptamers, and may open an approach to screening aptamer libraries in vivo.

Biophysical characterization of DNA aptamer interactions with vascular endothelial growth factor.

The binding of a DNA aptamer (5'-CCGTCTTCCAGACAAGAGTGCAGGG-3') to recombinant human vascular endothelial growth factor (VEGF(165)) was characterized using surface plasmon resonance (SPR), fluorescence anisotropy and isothermal titration calorimetry (ITC). Results from both fluorescence anisotropy and ITC indicated that a single aptamer molecule binds to each VEGF homodimer, unlike other VEGF inhibitors that exhibit 2(ligand):1(VEGF homodimer) stoichiometry. In addition, ITC revealed that the association of the aptamer to VEGF at 20 degrees C is enthalpically driven, with an unfavorable entropy contribution. SPR kinetic studies, with careful control of possible mass transfer effects, demonstrated that the aptamer binds to VEGF with an association rate constant k(on) = 4.79 +/- 0.03 x 10(4) M(-1) s(-1) and a dissociation rate constant k(off) = 5.21 +/- 0.02 x 10(-4) s(-1) at 25 degrees C. Key recognition hot-spots were determined by a combination of aptamer sequence substitutions, truncations, and extensions. Most single-nucleotide substitutions, particularly within an mfold-predicted stem, suppress binding, whereas those within a predicted loop have a minimal effect. The 5'-end of the aptamer plays a key role in VEGF recognition, as a single-nucleotide truncation abolished VEGF binding. Conversely, an 11-fold increase in the association rate (and affinity) is observed with a single cytosine nucleotide extension, due to pairing of the 3'-GGG with 5'-CCC in the extended aptamer. Our approach effectively maps the secondary structural elements in the free aptamer, which present the unpaired interface for high affinity VEGF recognition. These data demonstrate that a directed binding analysis can be used in concert with library screening to characterize and improve aptamer/ligand recognition.

Water-elutability of nucleic acids from metal-chelate affinity adsorbents: enhancement by control of surface charge density.

Immobilized metal affinity chromatography (IMAC) is widely used for purification of proteins, especially "hexahistidine-tagged" recombinant proteins. We previously demonstrated the application of IMAC to selective capture of nucleic acids, including RNA, selectively-denatured genomic DNA, and PCR primers through interactions with purine bases exposed in single-stranded regions. We also found that the binding affinity of nucleic acids for IMAC adsorbents can be increased several-fold by addition of 20 volume% of neutral additives such as ethanol or DMSO. In the present work, it is demonstrated that bound nucleic acids can be effectively eluted with water instead of the usual imidazole-containing competitive eluants, when the surface density of negative charges is enhanced by operation at alkaline pH, or by deliberate metal-underloading of the anionic chelating ligands. With enhanced negative surface charge density, nucleic acid adsorption can be made strongly dependent on the presence of adsorption-promoting additives and/or repulsion-shielding salts, and removal of these induces elution. Complete water-elutability is demonstrated for baker's yeast RNA bound to 10% Cu(II)- underloaded IDA Chelating Sepharose in a binding buffer of 20 mM HEPES, 240 mM NaCl, pH 7. Water elutability will significantly enhance the utility of IMAC in nucleic acid separations.

Neutral additives enhance the metal-chelate affinity adsorption of nucleic acids: role of water activity.

Immobilized metal-chelate affinity chromatography has been widely used in the purification of proteins, and we have recently found that it can also be applied to purification of nucleic acids through interactions involving exposed bases, especially purines. Here we report that the inclusion of moderate quantities of neutral solutes in the buffer substantially enhances the binding affinity of nucleic acids for immobilized metal-chelate affinity adsorbents. Addition of 20% (v/v) of solutes such as ethanol, methanol, isopropanol, n-propanol, and dimethyl sulfoxide enhances the initial affinity of binding of total yeast RNA by 4.4-, 3.8-, 3.7-, 3.0-, and 2.8-fold, respectively for Cu(II)-iminodiacetic acid (IDA) agarose adsorbent, and the weaker adsorption by Cu(II)-nitrilotriacetic acid (NTA) agarose was even more strongly enhanced. The adsorption affinities of the smaller oligodeoxynucleotide molecules A20, G20, C20 and T20 also increase with the addition of ethanol, suggesting that the effect is not significantly mediated by conformational changes. Binding enhancement generally correlates with reduction of water activity by the various solutes, as predicted by several models of solution thermodynamics, consistent with an entropic contribution by displacement of waters from the metal-chelate. Interestingly, the enhancement was not seen with the proteins bovine serum albumin and lysozyme.