Rethinking PICO in the Machine Learning Era: ML-PICO

Background  Machine learning (ML) has captured the attention of many clinicians who may not have formal training in this area but are otherwise increasingly exposed to ML literature that may be relevant to their clinical specialties. ML papers that follow an outcomes-based research format can be assessed using clinical research appraisal frameworks such as PICO (Population, Intervention, Comparison, Outcome). However, the PICO frameworks strain when applied to ML papers that create new ML models, which are akin to diagnostic tests. There is a need for a new framework to help assess such papers. Objective  We propose a new framework to help clinicians systematically read and evaluate medical ML papers whose aim is to create a new ML model: ML-PICO (Machine Learning, Population, Identification, Crosscheck, Outcomes). We describe how the ML-PICO framework can be applied toward appraising literature describing ML models for health care. Conclusion  The relevance of ML to practitioners of clinical medicine is steadily increasing with a growing body of literature. Therefore, it is increasingly important for clinicians to be familiar with how to assess and best utilize these tools. In this paper we have described a practical framework on how to read ML papers that create a new ML model (or diagnostic test): ML-PICO. We hope that this can be used by clinicians to better evaluate the quality and utility of ML papers.     

IFN-alpha1,8 inhibits tumor-induced angiogenesis in murine angiosarcomas.

Interferon-alpha (IFN-alpha) has proved effective in the treatment of hemangiomas, hemangioblastomas, and Kaposi's sarcoma. To investigate the ability of IFNs to inhibit angiosarcoma, we used two transformed murine endothelial cell lines that form angiosarcomas in vivo. SVR and MS1-VEGF cell lines express oncogenic H-ras or vascular endothelial growth factor (VEGF), respectively. IFN-alpha1,8, which is active against murine and human cells, inhibited SVR and MS1-VEGF proliferation in vitro by 40% at 10(3) U/mL (p = 0.028). In vivo, IFN-alpha1,8 inhibited SVR tumor volume by 71% (p = 0.047) and MS1-VEGF volume by 79% (p = 0.003). Tumor-induced angiogenesis was decreased in SVR tumors by 52% (p = 0.005) and in MS1-VEGF tumors by 58% (p = 0.001). Sera from IFN-alpha1,8-treated mice bearing either SVR or MS1-VEGF tumors demonstrated a 5-fold increase in IP-10/CXCL10 (p = 0.001), an IFN-induced antiangiogenic protein. Both recombinant IP-10 and IFN-alpha1,8 inhibited human umbilical vein endothelial cell (HUVEC) vessel formation in the fibrin gel assay, a three-dimensional culture model of angiogenesis, by 56% at 25 ng/mL and 50% at 1.2 ng/mL, respectively (p < 0.001). An IP-10 blocking antibody restored vessel formation to 80% of untreated controls (p = 0.001). Given the magnitude of the in vivo response, these data suggested that the antitumor effects of IFN-alpha1,8 were likely mediated through angiogenesis inhibition rather than solely by direct inhibition of tumor cell proliferation.     

Probable gas embolism during operative hysteroscopy caused by products of combustion

PURPOSE: Gas embolism is a rare but well documented entity during operative hysteroscopy, with an incidence of 10-50%. Catastrophic outcomes occur at a rate of three in 17,000 procedures. The purpose of this report is to present a non-fatal case of gas embolism probably caused by the gaseous products of combustion. CLINICAL FEATURES: A 50-yr-old woman with a history of menorrhagia was scheduled for hysteroscopy and endometrial ablation and polypectomy. Fifteen minutes into the procedure, with the patient in lithotomy position, 20 degree head down tilt, and breathing spontaneously, a sudden oxygen desaturation occurred from 97% to 87%. The patient's end-tidal carbon dioxide dropped from 46 mmHg to 27 mmHg. The patient's breathing pattern remained normal, respiratory rate remained 11-12 breaths x min(-1) but amplitude of the reservoir bag movement was increased. Cardiovascular variables remained stable. She responded rapidly to 100% oxygen and made an uneventful recovery. Having ruled out other possible causes, we concluded gas embolism was responsible for the fall in oxygen saturation and end-tidal CO(2). CONCLUSION: With all the precautions in place to minimize the likelihood of fluid overload and ambient air embolism occurring, we surmised that products of combustion were the cause of the gas embolism. During endometrial ablation, gaseous products of combustion, mainly carbon dioxide, accumulate. The gases may then contribute to the rise in uterine pressure that occurs as irrigation fluid enters the uterus and this rise in pressure in turn encourages passage of gas into the open venous sinuses.     

Temporal summation of repetitive click stimuli

Monaural loudness balances were performed by eight normal-hearing subjects to determine the effect of click repetition rate on loudness sensation. Click trains of 500 msec duration were matched in loudness to a standard 500 msec 1000 Hz tone burst presented at three reference loudness levels (70, 80, and 90 phons). Click trains were presented at repetition rates of 11, 31, 51, and 91 clicks per sec. It was found that click trains at faster repetition rates required lower intensities for judgments of equal loudness sensation. This finding was attributed to the process of temporal loudness summation. The magnitude and nature of the temporal summation process as well as the influence of the reference loudness level are discussed.     

Cough and paradoxical vocal fold motion

OBJECTIVES: The differential diagnosis and treatment of patients with chronic cough, paradoxical vocal fold motion, and disordered breathing can be a challenge to most practicing otolaryngologists. Tracheobronchial (ie, asthma, bronchitis, and tracheal stenosis), laryngeal (ie, vocal fold paralysis and neoplasms), and rhinologic (ie, allergies and rhinosinusitis) etiologies are commonly diagnosed and treated effectively. However, occasionally one is faced with patients who are refractory to medical treatment and have no obvious rhinologic, laryngeal or pulmonary cause. STUDY DESIGN AND SETTING: We conducted a review of the literature. METHODS: We present a thorough review of the current medical literature exploring the complex neurologic mechanisms involved in the production of cough and the relationship between gastroesophageal reflux disease, vagal neurapathy, and paradoxical vocal fold motion. RESULTS: The diagnosis and successful treatment of chronic cough can be complex. It requires a thorough understanding of the neurologic mechanisms behind cough excitation and suppression. Successful treatment strategies include aggressive management of the patient's reactive airway disease, gastroesophageal reflux disease, and, in select cases, paradoxical vocal fold motion. This may involve a well-coordinated effort among pulmonologists, otolaryngologists, gastroenterologists, and speech pathologists. CONCLUSION: Gastroesophageal reflux disease, vagal neuropathy, and paradoxical vocal fold motion are additional causes of chronic cough and disordered breathing that need to be considered, in the absence of obvious laryngotracheal and/or rhinologic pathology. A high index of suspicion is essential in making the diagnosis and formulating an effective multidisciplinary treatment plan for these patients.