Yes, you can use Random Forest (RF) technique, that is used in medical bioinformatics including a number of decision trees that incorporates feature selection and acts their interactions naturally in the learning process. RF technique provides a unique combination of prediction accuracy and model interpretability among popular machine learning methods. The random sampling and ensemble strategies utilized in RF enable it to achieve accurate predictions as well as better generalizations.

My suggestion is you can use hematological, chemistry, urine markers with some vital and daily life markers under RF technique. Sampling will be tested 3 times during 9 months of gestation. First sample as a baseline of 1st month, then 3rd month, 3rd sample at the completion of 2nd trimester and 4th sample on 3rd-trimester completion. You will find different disease predictions throughout 4 sampling cycles.

The parameters are:  

Hematological parameters

Hemoglobin

Hematocrit

RBCs count

WBCs count

Platelets count

MCV

MCH

MCHC

RDW

Neutrophils %

Lymphocytes %

Eosinophils %

Monocytes %

Clinical Chemistry parameters

Plasma Glucose Fasting

Plasma Glucose Random

Glucose Challenge test

Serum calcium

Total protein

Serum Phosphorus

ALT

AST

Urine Examination parameters

Urine detailed report

Urine protein (albumin)

Other factors

Temperature

weight

Blood pressure

Pulse rate

Smoking

Alcohol intake

Go ahead if you agree for my suggestions.  

There is no improved outcomes by over predicting risk of severe morbidity and mortality.  Instead spend time doing nutritional counseling and make sure the woman has adequate micro and macro nutrients.  If  her blood pressure is high, have her also eat a whole head of fresh garlic per day and it will take it down to normal.     

Dear Dr Juan Carlos Martinez Santos

The question is: Which state you want to predict?

I think, that very good software/tools you can find on the website of Fetal Medicine Foundation.

Best regards