Pregnancy of unknown location (PUL) is a temporary pathologic or physiologic phenomenon of early pregnancy that requires follow up to determine the final pregnancy outcome. Evidence indicated that PUL patients suffer a remarkably higher rate of adverse pregnancy outcomes, represented by ectopic gestation and early pregnancy loss, than the general population. In the past few decades, discussion about PUL has never stopped, and a variety of markers have been widely investigated for the early and accurate evaluation of PUL, including serum biomarkers, ultrasound imaging features, multivariate analysis, and the diagnosis of ectopic pregnancy based on risk stratification. So far, machine learning (ML) methods represented by M4 and M6 logistic regression have gained a level of recognition and are continually improving. Nevertheless, the heterogeneity of PUL markers, mainly caused by the limited sample size, the differences in population and technical maturity, etc., have hampered the management of PUL. With the advancement of multidisciplinary integration and cutting-edge technologies (e.g. artificial intelligence, prediction model development, and telemedicine), novel markers, and strategies for the management of PUL are expected to be developed. In this review, we summarize both conventional and novel markers (represented by artificial intelligence) for PUL assessment and management, investigate their advancements, limitations and challenges, and propose insights on future research direction and clinical application.

BACKGROUND: There is no international consensus on how to manage women with a pregnancy of unknown location (PUL).
OBJECTIVE: To present a systematic quantitative review summarising the evidence related to management protocols for PUL.
METHODS: MEDLINE, COCHRANE and DARE databases were searched from 1 January 1984 to 31 January 2017. The primary outcome was accurate risk prediction of women initially diagnosed with a PUL having an ectopic pregnancy (high risk) as opposed to either a failed PUL or intrauterine pregnancy (low risk).
METHODS: All studies written in the English language, which were not case reports or series that assessed women classified as having a PUL at initial ultrasound.
METHODS: Forty-three studies were included. QUADAS-2 criteria were used to assess the risk of bias. We used a novel, linear mixed-effects model and constructed summary receiver operating characteristic curves for the thresholds of interest.
RESULTS: There was a high risk of differential verification bias in most studies. Meta-analyses of accuracy were performed on (i) single human chorionic gonadotrophin (hCG) cut-off levels, (ii) hCG ratio (hCG at 48 hours/initial hCG), (iii) single progesterone cut-off levels and (iv) the \'M4 model\' (a logistic regression model based on the initial hCG and hCG ratio). For predicting an ectopic pregnancy, the areas under the curves (95% CI) for these four management protocols were as follows: (i) 0.42 (0.00-0.99), (ii) 0.69 (0.57-0.78), (iii) 0.69 (0.54-0.81) and (iv) 0.87 (0.83-0.91), respectively.
CONCLUSIONS: The M4 model was the best available method for predicting a final outcome of ectopic pregnancy. Developing and validating risk prediction models may optimise the management of PUL.
CONCLUSIONS: Pregnancy of unknown location meta-analysis: M4 model has best test performance to predict ectopic pregnancy.