References:

Health Information Management Case Studies Second Edition AHIMA by Dianna M. Foley

Health Information Management Technology an Applied Approah Sixth Edition AHIMA by Nanette Sayles and Leslie Gordon

Extended Cases 7.2 Coding productivity process

Case Study: if not from your own thinking I will reject

Aspen Hills Hospital's HIM director, Sabrina Cavanaugh is meeting with her coding supervisor, Patricia Reynolds, who has been concerned about inpatient coding productivity. Since ICD-10 implementation, productivity numbers have fallen and never rebounded like Patricia expected they would as coders became more familiar with the new classification system. Outsourcing costs as well as the discharged not final billed (DNFB) totals have been rising. There is pressure from administration to lower or eliminate the costs associated with outsourcing altogether. Currently, the outsourcing productivity standard is two charts per hour and rate of pay is $35.00 per hour. Inpatient discharges are relatively stable pre- and post-ICD-10 implementation, running between 60 to 70 discharges per day. Prior to ICD-10, there was no outsourcing, with seasonal increases in discharge volumes handled by overtime.

The following are coding staff details; productivity is calculated per hour.

The full coding staff receive monthly education and, in general, have very good coding quality scores. All credentialed coders maintain their certifications with CEUs from various webinars and seminars they attend. The group works remotely, but technology is challenging for Kathy, Robert, and Charles. They often have to call IT for support with issues that the other coders can troubleshoot themselves. This has strained the relationship between IT and HIM. IT sometimes must spend a significant amount of time fixing issues for the remote coders. Coders feel that IT staff do not recognize the importance of coding to the bottom line of the hospital, especially when their requests for help are not answered in a timely manner or taken seriously. Inpatient and outpatient coders are not cross-trained, but all outpatient coders can code all varieties of outpatient records, including ancillary accounts, ERs, outpatient surgeries, and observations. There is little cohesiveness between the inpatient and outpatient coders. There has been little turnover in the coding staff, but one coder is approaching retirement, and another was hired just two years ago, right out of college. They have 30 and 2 years of experience respectively, while the other coders average 12 years of experience. Coding staff are insistent that the current productivity standards are unreasonable. One of the things they take issue with is that the standards are built on the assumption of a full 8 hours a day spent coding. Staff know that they spend time throughout their day on other activities not just coding.

1. One of the first things that Sabrina asks Patricia to do is a work measurement study. Using the information below, finish building the table and propose a new standard for inpatient productive time (round to whole number). All three coders worked 80 hours in this two-week period for the work-study.

2. Using the coding staff data supplied previously, make a bar graph to illustrate the pre and post ICD- 10 implementation productivity for both individual inpatient coder and for the inpatient coding team as a whole.

Patricia thinks implementing a computer-assisted coding (CAC) program might help with the productivity issues. She is aware that the number of CAC vendors has grown over the past three years, including 3M, Optum, and others. In her research, she finds that the CAC software may need a specific interface in order to function. In addition, the cost of CAC software varies depending on the information system to which it must tie in and the number of users of the application. Trained coders should be used in conjunction with a CAC program to review the proposed codes and make a determination on their appropriateness. Vendors claim that some of the cost of a CAC can be recouped with savings achieved through increased coder productivity. Patricia's investigation shows that increases in coder productivity can range from 20 to 35 percent. The requests for proposal have identified one vendor that stands out. The entire cost of the implementation would be $80,000, and they are standing behind a 25 percent productivity increase. Additional benefits may be realized in reduction of time spend on denials and queries and an increase in the case-mix index.

3. Make a SWOT analysis regarding the purchase of a CAC application.

4. Based on the time study results for actual time that inpatient coders spend coding, present the figures for the number of charts coded daily and biweekly for the coding team overall. Be sure to present the numbers for pre and post ICD- 10 implementation as well as if a 25 percent productivity increase was realized with a CAC implementation. Provide calculations to nearest whole number. Then forecast the annual savings on outsourcing expenditures that would occur with a 25 percent productivity increase.

5. Sabrina know that the organization requires at least a 33 percent return on investment for major purchases such as CAC software. Compile the data necessary to arrive at the return on investment and forecast if Sabrina will be able to receive the necessary funds for the purchase.

6. Make a GANTT chart to manage the project based on the following information.

The decision was made to proceed with the CAC implementation and Sabrina delegated the management of this project to Patricia. Patricia recognized the importance of balancing a smooth implementation process with the need to get this project completed in a quick manner. Therefore, she developed an aggressive timeline for the project from the contract through the go-live scheduled for 2/24/20. The kick-off for the project will begin with the contract completion process. At A pen Hills, the department director and the CEO must both sign off on the contract. Patricia sets the start date as 1/20/2020. She allows 3 days for Sabrina to sign the contract, and another 8 days for the CEO, Judith Adler to sign and return the contract. Unfortunately, Sabrina has the nu the week the contract is to be signed and does not come into work. On Thursday, 1/23/2020, she calls Patricia to touch ba e and realizes that the project is going to be a week behind schedule if she does not sign the contract until she returns to work next week. She has Patricia fax her the contract, which she immediately signs and faxes back. Patricia sends the contract to the CEO the next day and gets it back one day sooner than the originally planned date of 1/30/2020. On 1/30/2020, the signed contract is sent to the vendor for signatures, which Patricia had originally allowed to be completed by 2/5/2020 but actually took until 2/7/2020. When Patricia realized the contract was not going to be finished on time, he needed to reschedule the installation of the software, which had been planned for 2/7/2020. This had to be pushed back a week. until 2/14/2020, because of the vendor's other commitments. Similarly, the testing phase was rescheduled as well, moving from 2/10 through 2/11 to 2/15 through 2/16. Since Patricia was responsible for the testing phase, she decided to work over the weekend to help move the project closer to being on track. Patricia decided that in addition to herself, they would train one inpatient and one outpatient coder as super users. She chose the most experienced coder from each group who were technology savvy to train. She originally allotted three days for this training, from 2/12 through 2/14. As it turned out, the training bad to be delayed to a 2/17 start date because the installation and testing ran over. Luckily, the vendor was able to accommodate her request. Their super user training finished a day earlier than planned as they all caught on very quickly to the use of the software. By virtue of the ease-of-use of the software, Patricia felt that the rest of the coders could be trained in three days instead of five as originally planned. Therefore, the training could be moved from 2/17 through 2/21 to 2/19 through 2/21, thus preserving the go-Iive date of 2/24, which went off without a hitch.(a) Describe an efficient algorithm to determine whether two finite line segments in a plane intersect. You may assume that the end points of each line are given as x-y coordinates. [8 marks] (b) Describe in detail an efficient algorithm to find the convex hull of a set of points lying on a plane. Show that the complexity of the Graham scan used in the algorithm is O(n) and that the algorithm as a whole has complexity O(n log n). [8 marks] (c) Discuss how it is possible to eliminate many of the points before the convex hull algorithm is entered. [4 marks] 2 Computer Design The ARM processor allows the second operand to be shifted by an arbitrary amount. In order to improve the performance, a six-stage pipeline is proposed with the following stages: instruction decode and shift execute memory register fetch register fetch operand 2 access write back (a) What are control hazards and how could they be resolved in the above pipeline? [4 marks] (b) What are data hazards and how could they be resolved in the above pipeline? [4 marks] (c) What are feed-forward paths and where could they be added to the above pipeline to improve performance? [6 marks] (d) Why might a branch instruction result in pipeline bubbles and how many bubbles will appear in the above pipeline as a result of taking a branch instruction? [6 marks] 2 CST.2004.13.3 3 Digital Communication I (a) Define the terms latency and capacity as applied to communication channels. [2 marks] (b) Is there a strict relation between the two? [1 mark] (c) Show how the latency of a channel can have a direct effect on the capacity of a higher-layer channel which uses it. [10 marks] (d) How can the capacity of the higher-layer channel be improved (keeping the characteristics of the underlying channel unchanged)? [4 marks] (e) In what circumstances might these improvements have only limited benefit? [3 marks] 3 [TURN OVER CST.2004.13.4 4 Distributed Systems A network-based service manages persistent objects. The service must enforce an access control policy to protect the objects.

Network Y (WAN) requires 2 Usable host IP addresses. (Use the next available subnet) Network Z (WAN) requires 2 Usable host IP addresses. (Use the next available subnet) Subnet Table: Subnet Description Network Address First Host Address Second Host Address Last Host Address Broadcast Address Subnet Mask Network P Network Network R Network X NA Network Y NA Network Z NA Part 2: Complete the Addressing Table Document the addressing scheme and complete the addressing table. a. Assign the Last host IP address from Network P subnet to G0/1 interface of Ri. b. Assign the First host IP address from Network P subnet to Sl. c. Assign the Second host IP address from Network P subnet to PCI. d. Assign the Last host IP address from Network Q subnet to GO/1 interface of R2. e. Assign the First host IP address from Network Q subnet to S2. f. Assign the Second host IP address from Network Q subnet to PC2. g. Assign the Last host IP address from Network R subnet to G0/1 interface of R3. h. Assign the First host IP address from Network R subnet to S3. i. Assign the Second host IP address from Network R subnet to PC3. j. Assign the First host IP address from Network X subnet to S0/1/0 interface of Ri. k. Assign the Last host IP address from Network X subnet to S0/1/1 interface of R2. 1. Assign the First host IP address from Network Y subnet to S0/1/0 interface of R2. m. Assign the Last host IP address from Network Y subnet to S0/1/1 interface of R3. n. Assign the First host IP address from Network Z subnet to S0/1/0 interface of R3. 0. Assign the Last host IP address from Network Z subnet to S0/1/1 interface of Ri. (b) Give a grammar which is ambiguous. [2 marks] (c) Give a grammar which is not a regular grammar but which generates a regular language containing an infinite number of strings. [2 marks] (d) Is it possible to a grammar which generates the strings {aa, aaa, aaaaa, . . . , ap , . . .} where p is prime? (A general argument for or against suffices.) [2 marks] (e) It is desired to construct a simple "pocket-calculator" program using yacc and lex (or other similar automated tools of your choice) which can parse strings such as "1+(10-5-3)*5+2=" and print the result, 13 in this case. Outline the overall structure of your program components. Give full details of the input to yacc and lex (or equivalent). (Precise syntactic details are not important, but your answer should show an understanding of the principles involved.) [8 marks] 4 CST.2005.11.5 4 Data Structures and Algorithms (a) A closed hash table is one in which the overflow chains of key-value pairs are held within the table itself. Carefully describe how the closed hash table mechanism works for both insertion and lookup. [6 marks] (b) Assume that the initial probe is p0 = Hash1(key) mod B and the secondary probes are pi , i = 1 . . . B 1. Discuss the relative merits of the following schemes for choosing the secondary probes. (i) pi = (p0 + i) mod B (ii) pi = (p0 + 13 i) mod B (iii) pi = (p0 + 13 i + 17 i i) mod B (iv) pi = (p0 + Hash2(key) i + 17 i i) mod B You may assume that all the arithmetic is unsigned. [8 marks] (c) Carefully describe a mechanism for deleting key-value pairs from a closed hash table. [6 marks] 5 [TURN OVER CST.2005.11.6 5 Artificial Intelligence I A perceptron computes the function h(x) = sgn(wT x + w0) where sgn(x) = +1 if x 0 and sgn(x) = 1 otherwise. The primal perceptron algorithm is as follows: do { for (each example in s) { if (yi(wT xi + w0) 0) { w = w + yixi w0 = w0 + yiR2 } } } while (mistakes are made in the for loop) where is a positive real, R = maxi ||xi || and w and w0 are initialised to be the zero vector and 0 respectively, and where s = ((x1, y1),(x2, y2), . . . ,(xm, ym)) with yi {+1, 1} is a training sequence. (a) Derive the dual form of the perceptron algorithm and state the corresponding alternative representation for w. [6 marks] (b) Explain how the perceptron may be applied to problems that are not linearly separable by introducing basis functions. [4 marks] (c) Give a definition of a kernel. [3 marks] (d) Explain how the use of a suitable kernel in conjunction with the dual form of the perceptron algorithm can be advantageous compared with the direct use of basis functions and the primal perceptron algorithm. [7 marks] 6 CST.2005.11.7 6 Operating System Foundations (a) A system has paging hardware but no segmentation hardware. Discuss the likely structure of a process page table. What information would you expect to be held in a page table entry? [5 marks] (b) Describe the operation of the following hardware support options for paging: (i) A pair of processor registers. PTBR holds the address of the base of the page-table of the current process. PTLR holds the size of the page table in bytes. [5 marks] (ii) A TLB (Translation Lookaside Buffer). [10 marks] 7 [TURN OVER CST.2005.11.8 7 Continuous Mathematics (a) Let f(x) be a periodic function of period 2. Give expressions for the Fourier coefficients ar (r = 0, 1, . . .) and br (r = 1, 2, . . .) of f(x) where a0 2 + X r=1 (ar cos rx + br sin rx) is the Fourier series representation of f(x). [2 marks] (b) Show that the Fourier series in part (a) can also be written as a complex Fourier series rX= r= cre irx by deriving expressions for the complex Fourier coefficients cr (r = 0, 1, 2, . . .) in terms of ar and br. [3 marks] (c) Use your expressions for ar and br in part (a) and for cr in part (b) to show that cr = 1 2 Z f(x)e irxdx (r = 0, 1, 2, . . .). [3 marks] (d) Show that the complex Fourier coefficients of f(x ) (where is a constant) are given by e ircr (r = 0, 1, 2, . . .). [6 marks] (e) Suppose that g(x) is another periodic function of period 2 with complex Fourier coefficients dr (r = 0, 1, 2, . . .) and define h(x) by h(x) = 1 2 Z f(x y)g(y)dy . Show that h(x) is a periodic function of period 2 and that its complex Fourier coefficients are given by hr = crdr (r = 0, 1, 2, . . .). [6 marks] [You may assume that the periodic functions in this question satisfy the Dirichlet conditions. Euler's equation may be used without proof but should be stated precisely.] 8 CST.2005.11.9 8 Numerical Analysis I (a) Define absolute error and relative error. How are these related? Explain briefly the term loss of significance. [3 marks] (b) An algorithm is required for solution of ax2 +bx+c = 0 where b > 0. Describe how loss of significance can occur in the formula x = b b 2 4ac 2a and derive an alternative formula for such a case. Illustrate your answer by applying it to the case a = 30, b = 3000, c = 1 on a decimal machine with only 5 significant digits available. [You should assume for the purposes of calculation that b 2 4ac evaluates to b (2ac/b), correctly rounded.] [10 marks] (c) The series cos x = 1 x 2 2! + x 4 4! x 6 6! + . . . is to be summed by taking terms in order, left to right, using only p decimal digits of precision until additional terms are negligible. If x = 6 find the largest term of the series and hence, assuming cos 6 ' 1, estimate roughly how many decimal digits of accuracy will be lost in the process. [7 marks] 9 [TURN OVER CST.2005.11.10 9 Mathematics for Computation Theory (a) Let M be an n-state deterministic finite automaton over the finite alphabet S. Write l(w) for the length of words w S . Suppose that M accepts the word x S , where l(x) > n. (i) Show that x is a concatenation of words uvw, where l(uv) 6 n, l(v) > 1, and M accepts the word zk = uvkw for all natural numbers k > 0. [8 marks] (ii) Hence show that if M accepts some word y S , it must accept some word z S such that l(z) < n; and that M accepts an infinite set of words if and only if it accepts some word x S such that n 6 l(x) < 2n. [5 marks] (b) Let S = {a, b} be an alphabet of two symbols. Explain whether each of the following languages over S is regular: (i) L1 = {uv | u, v S , l(v) = 2.l(u)} [3 marks] (ii) L2 = {ww | w S } [4 marks] 10 CST.2005.11.11 10 Computation Theory (a) What does it mean for a subset S of the set N of natural numbers to be register machine decidable? [3 marks] (b) For each e N, let e Pfn(N, N) denote the partial function computed by the register machine program with index e. Let e0 N be an index for the totally undefined partial function (so that e0 (x), for all x N). Suppose that a total function f Fun(N, N) is extensional, in the sense that for all e, e0 N, f(e) = f(e 0 ) if e and e 0 are equal partial functions. Suppose also that the set Sf = {x N | f(x) = f(e0)} is not the whole of N, so that for some e1 N, f(e1) 6= f(e0). (i) If membership of Sf were decided by a register machine M, show informally how to construct from M a register machine M0 that, started with R1 = e and R2 = n (any e, n N) always halts, with R0 = 0 if e(n) and with R0 = 1 if e(n). Make clear in your argument where you use the fact that f is extensional. [Hint: For each e, n N consider the index i(e, n) N of the register machine that inputs x, computes e(n) and if that computation halts, then computes e1 (x).] [14 marks] (ii) Deduce that if f is extensional, then Sf is either the whole of N, or not decidable. [3 marks] 11 Software Engineering and Design Imagine that you are the software designer responsible for a system that manages the shutdown of a nuclear power station. In normal shutdown mode fuel rods are withdrawn immediately, but steam generation is stopped only after turbine speed has dropped to one quarter. However, in emergency shutdown mode steam generation must be stopped immediately. (a) Sketch initial UML diagrams showing important aspects of your design, including a use case diagram, a class diagram, a sequence diagram and a statechart diagram. [4 marks each] (b) What international quality standards should be applied in managing these and subsequent design documents? What procedures would result?