Dilatation and Curettage

This research covered a sample of 800 drivers (of category D and C) aged 23- 75 who were referred to Imam Sajjad Centre for drug Addiction Diagnosis. Convenience sampling method was used and it was done at two stages. Questionnaires were distributed to the drivers who had been referred to check their addiction after ensuring that all the drivers regularly used large vehicles. We also checked that all questions were answered. Then, all the drivers were selected to answer all interview items. Inclusion criterion was as follows: Drivers (of category D and C) who were referred to Imam Sajjad Centre; Exclusion criteria were as follows: Female drivers, and illiterate or uneducated drivers who could not understand the questions and refused to complete the questionnaires.
Manchester Driving Behavior Questionnaire (MDBQ):This scale was adjusted and compiled by Rissen et al. in the psychology department of ‎Manchester University (19 ). It is based on the idea that errors and violations have different ‎psychological reasons and correction methods; hence, they should be discriminated by ‎researchers. Today, MDBQ has been changed into a popular instrument for assessing driving ‎behaviors. This scale has 50 questions with Likert range from 0 to 5. Questions have two ‎different aspects. One aspect is about the kind of behavior, and another relates to amount of ‎risk posed to other drivers. Abnormal behaviors are as follows: Lapse errors, slips, deliberate ‎violation and unintentional violation. These behaviors are classified as follows:
1.    Behaviors that pose no risk to others, and just give a feeling of comfort (low risk ‎probability)‎
2.    Behaviors that are likely to put others at risk (moderate risk probability)‎
3.    Behaviors that certainly put others at risk (high risk probability)‎
MDBQ has acceptable psychometric properties. Parker and Reason (20 ) have obtained a ‎correlation coefficient of 0.81 for errors and 0. 75 for violation in reliability research ‎for 80 drivers with a seven-week interval. ‎
For data analysis, we used factor analysis (to analyze construct validity), internal consistency ‎‎(Chronbach’α), split half, and test-retest, respectively. Less than 0.05 were considered to be ‎statistically significant.‎

The previous publications have derived the concepts of TSC and TSE, i.e., the exposures of one or multiple compounds that result in tumor stasis [15 (link), 16 (link), 20 (link)]. The transition from TSC to TSE was made to include exposure metrics other than plasma concentrations, in particular doses of ionizing radiation [15 (link)]. For combinations of radiation and radiosensitizer, the associated TSE curve consists of all pairs of radiation dose and plasma concentration, such that exposures above the curve will lead to tumor regression. One can derive a TSE curve based on the long-term treatment effect using the model in Eq. 2. The TSE curve is derived from the equation for $V_1$ by considering for which total radiation doses and radiosensitizer concentrations the growth rate becomes equal to the natural death rate. Any exposure combination above this level will result in a negative net growth rate and, therefore, tumor shrinkage. In these calculations, the short-term radiation effect can be ignored, since it only has a temporary effect on tumor volume. From Eq. 2, the growth and kill rates will be equal when $$k_{\text{net}}\text{:=}{k}_{\text{g}}exp(-\alpha {\text{IR}}_{\text{Tot}})-k_k=0,$$ where $k_{\mathit{net}}$ is the net growth rate. Moreover, the (effective) total radiation dose, after an effective increase due to radiosensitizing treatment is accounted for, is given by $\left(1+aC\right)D$ , where total radiation dose and radiosensitizer concentration are denoted D and $C$ , respectively. Inserting this into Eq. 9 yields the following: $$k_{\text{g}}exp(-\alpha \left(1+aC\right)D)-k_{\text{k}}=0.$$
Equation 10 describes a curve in the plane with the plasma concentration $C$ along the horizontal axis, and the total radiation dose and the total radiation dose $D$ along the vertical axis. Equation 10 can be solved for $D$ to obtain the following: $$\frac{D=log(k_{\text{g}}/k_{\text{k}})}{\alpha (1+aC)}.$$
Thus, for each value of the plasma exposure of the radiosensitizer $C$ , the right-hand side of Eq. 11 gives the necessary total radiation dose $D$ , such that the tumor will eventually be eradicated. Equation 11 can be viewed as a function: $$D=f(\theta ;C),$$ where $\theta$ is the vector of parameters (here $k_{\text{g}},k_{\text{k}},\alpha ,\text{and}a$ ). The graph of this function will be the TSE curve. Alternatively, one can solve for $C$ in Eq. 11 to obtain $$C=\frac{log(k_{\text{g}}/k_{\text{k}})}{a\alpha D}-\frac{1}{a},$$ which describes the same curve as Eq. 11. Similarly, Eq. 13 can be viewed as a function $C=g(\theta ;D)$ that, for every radiation dose, $D$ determines the corresponding radiosensitizer concentration $C$ , such that the combination will lead to tumor eradication.

The setting for this research was the Western Cape province where existing public–private contracting for caesarean delivery services was occurring due to human resource shortages in rural district hospitals. Five rural district hospitals within one rural district were chosen following engagement with provincial managers and obstetric clinical managers.
In SA, women with low-risk pregnancies receive antenatal care at primary care clinics and community health centres. District hospitals provide level 1 (generalist) services to inpatients and outpatients including obstetric care for women with low-risk pregnancies. District hospitals have between 30 and 200 beds, a 24-hour emergency service and an operating theatre. Generalists (medical officers) provide the services together with nursing staff and allied health professionals; some district hospitals have specialist family physicians serving as clinical managers but there are no obstetric or anaesthetic specialists at district hospital level. Most district hospitals also have community service doctors. These are doctors who have completed a 2-year internship and are required to complete a further 1 year of community service.13 None of the five hospitals had newly qualified intern medical doctors who are generally not placed within district hospitals.
For obstetric services at district hospital level, normal vaginal deliveries are performed by midwives, assisted vaginal deliveries are performed by advanced midwives or medical officers and caesarean deliveries (surgery and anaesthesia) are performed by medical officers. Pregnant women with pre-existing morbidities such as diabetes, autoimmune disorders, thyroid disease, and cardiac disease or obstetric complications such as anticipated preterm delivery, suspected intrauterine growth restrictions, pre-eclampsia, placenta praevia, abruptio placentae, multiple pregnancy, two previous caesarean sections, body mass index over 35–40 kg/m², and severe anaemia are referred for delivery to a secondary or tertiary level hospital.
Public health facilities are permitted to contract the services of private providers where needed. There are three mechanisms by which private providers can be contracted to the public service: through a locum agency, through a sessional contract which is limited to a maximum of 39 hours per month or as a service provider in response to a tender for specific services. In all three contracting models, the remuneration is time based and not related to the number of patients or theatre cases performed. In the case of obstetric services, private providers are mainly used for theatre services either as a GP surgeon or GP anaesthetist to undertake caesarean deliveries or for obstetric surgery including ectopic pregnancy, termination of pregnancy and dilatation and curettage following spontaneous miscarriage. They may also be called for an assisted delivery if the establishment doctor is unable to manage a complicated delivery. For GPs contracted through a sessional contract, medicolegal indemnity is provided by the state but for those contracted as locums or through a service provider tender, they are required to have their own medicolegal indemnity cover. In these five hospitals, the private GPs did not have medical indemnity for private obstetric practice and only performed caesarean deliveries during their public sector contracted time.

The present study was conducted as a retrospective analysis at the Aga Khan University Hospital (AKUH), Karachi, Pakistan. Patients who underwent UAE from January 2010 to May 2020 were identified from the radiology database of the hospital. After excluding patients with known uterine tumours, retained products of conception, gestational trophoblastic disease and postpartum haemorrhage, 15 cases were identified in which UVAs were suspected clinically and/or on imaging.
A pre-structured pro forma was used to record patient demographics, including age, parity, pattern and volume of vaginal bleeding, history of uterine surgery or dilatation and curettage (D&C), time interval since the intervention, findings on imaging and angiography and patient outcome. The duration of hospital stays, post-procedure complication, follow-up ultrasound findings and post-embolisation fertility/pregnancy were also recorded. The patients’ imaging was reviewed using the picture archiving and communication system, Rogan Delft View Pro-X (Rogan-Delft BV, Veenendaal, Utrecht, Netherlands), while additional data was collected from the Health Information Management Services.
The pre-angiography imaging modality was chosen at the discretion of the referring physician and included ultrasonography with colour Doppler imaging, pelvic MRI and CT, either by itself or in combination. The referring physician decided on embolisation after consultation with the interventional radiologist who performed the embolisation procedures in the angiographic suite of the present hospital.
The patients underwent the procedure on a flat panel monoplane digital subtraction angiography machine, Axiom-Artis (Siemens Healthineers, Erlangen, Germany), under local anaesthesia. The femoral artery was punctured and a 4F vascular access sheath was inserted. A 4Fr Simmons 1 catheter (Cordis, Santa Clara, California, USA) or a Cobra 1 angiographic catheter (Cordis) was advanced over a 0.035-inch guide wire. An angiographic run was performed after selective catheterisation of the uterine artery. It was followed by super-selective cannulation using a Progreat® microcatheter (Terumo Interventional Systems, Tokyo, Japan), which was placed coaxially as near as possible to the feeder vessel. The embolisation materials used were polyvinyl alcohol particles (PVA) that were 355–500 μm in size, gel foam, glue and coil, either in combination or in isolation. In a few cases, the ovarian artery was also embolised. Clinical success was defined as the resolution of vaginal bleeding and/or abnormal imaging findings on post-embolisation follow-up.
The Statistical Package for the Social Sciences (SPSS), Version 20.0 (IBM Corp., Armonk, New York, USA) was used for statistical analysis. All quantitative data were expressed as mean ± standard deviation and qualitative data were expressed using frequencies and percentages. A descriptive analysis was conducted for all the variables, including the demographic and the other categorical variables and frequencies, proportions and percentages were reported.
The benefits and risks of the embolisation procedure were explained to the patients and it was performed only after their consent was obtained. Ethical approval for the study was obtained from the Ethical Review Committee of AKUH (ERC #2020-3690-10189).

Mathematically, a deep neural network (DNN) defines a mapping of the form $$\mathcal{F}:\mathit{x}\in {\mathbb{R}}^d⟹\mathit{y}=\mathcal{F}\left(\mathit{x}\right)\in {\mathbb{R}}^c,$$ where $d$ and $c$ are the dimensions of the input and output, respectively. Generally, a standard neural unit of a DNN receives an input $\mathit{x}\in {\mathbb{R}}^d$ and produces an output $\mathit{y}\in {\mathbb{R}}^m$ , i.e., $\mathit{y}=\sigma (\mathit{W}\mathit{x}+\mathit{b})$ with $\mathit{W}\in {\mathbb{R}}^{d\times m}$ and $\mathit{b}\in {\mathbb{R}}^m$ being weight matrix and bias vector, respectively. $\sigma \left(\cdot \right)$ , which is referred to as the activation function, is designed to add element-wise non-linearity to the model.
A DNN with $L$ hidden layers can be regarded as a nested composition of sequential standard neural units. For convenience, we denote the output of the DNN by $\mathit{y}(\mathit{x};\mathit{\theta })$ with $\mathit{\theta }$ standing for the set of all weights and biases. Specifically, the $j_{th}$ neuron in $\ell$ layer can be formulated as $$y_j^{\left[\ell \right]}=\sum _{k=1}^{N^{[\ell -1]}}{w}_{jk}^{\left[\ell \right]}\ast {\sigma }^{[\ell -1]}\left(y_k^{[\ell -1]}\right)+b_j^{\left[\ell \right]},$$ where $y_k^{[\ell -1]}$ represents the value of the $k_{th}$ neuron in the $\ell -1$ layer, $N^{[\ell -1]}$ represents the number of neurons in the $\ell -1$ layer, ${\sigma }^{[\ell -1]}$ is the activation function of the $\ell -1$ layer, $w_{jk}^{\left[\ell \right]}$ is the weight between the $k_{th}$ neuron in the $\ell -1$ layer and the $j_{th}$ neuron in the $\ell$ layer, and $b_j^{\left[\ell \right]}$ is the bias of the $j_{th}$ neuron in the $\ell$ layer.